.*)\.` to extract an appropriate file
Within `SaveCroppedObjects`, select `Default Output Folder sub-folder` and then **right-click** in the `sub-folder` text box and select `FileName`.
-```{figure} ./img/user-guide-save-cropped-objects-view-4.2.1.png
+```{figure} ../../img/cellprofiler-cell-crops-examples/user-guide-save-cropped-objects-view-4.2.1.png
---
name: SaveCroppedObjects-view-4.2.1
---
@@ -99,7 +99,7 @@ Right click within the `Sub-folder` text box and select `FileName`, as defined i
````
-```{figure} ./img/user-guide-save-cropped-objects-view.png
+```{figure} ../../img/cellprofiler-cell-crops-examples/user-guide-save-cropped-objects-view.png
---
name: SaveCroppedObjects-view
---
diff --git a/piximi-documentation/citing-piximi.md b/piximi-documentation/pages/technical/citing-piximi.md
similarity index 100%
rename from piximi-documentation/citing-piximi.md
rename to piximi-documentation/pages/technical/citing-piximi.md
diff --git a/piximi-documentation/example-datasets.md b/piximi-documentation/pages/technical/example-datasets.md
similarity index 90%
rename from piximi-documentation/example-datasets.md
rename to piximi-documentation/pages/technical/example-datasets.md
index cf6b664..260a8fe 100644
--- a/piximi-documentation/example-datasets.md
+++ b/piximi-documentation/pages/technical/example-datasets.md
@@ -9,7 +9,7 @@ If you use any of the example datasets in a publication, please also cite the or
`````{grid}
````{grid-item}
:columns: 3
-```{image} img/example_project_icons/mnistExampleProjectIcon.png
+```{image} ../../img/example_project_icons/mnistExampleProjectIcon.png
:alt: mnistExampleIcon
:align: left
```
@@ -26,12 +26,12 @@ MNIST citation doi: 10.1109/5.726791
````
`````
-## *C. elegans*
+## _C. elegans_
`````{grid}
````{grid-item}
:columns: 3
-```{image} img/example_project_icons/cElegansExampleProjectIcon.png
+```{image} ../../img/example_project_icons/cElegansExampleProjectIcon.png
:alt: cElegansExampleIcon
:align: left
```
@@ -55,7 +55,7 @@ The *C. elegans* data were provided by Javier Irazoqui as BBBC012 in the [Broad
`````{grid}
````{grid-item}
:columns: 3
-```{image} img/example_project_icons/humanU2OSCellsExampleProjectIcon.png
+```{image} ../../img/example_project_icons/humanU2OSCellsExampleProjectIcon.png
:alt: humanU2OSCellsExampleIcon
:align: left
```
@@ -77,7 +77,7 @@ U2OS citation doi: 10.1038/nmeth.2083
`````{grid}
````{grid-item}
:columns: 3
-```{image} img/example_project_icons/BBBC013ExampleProjectIcon.png
+```{image} ../../img/example_project_icons/BBBC013ExampleProjectIcon.png
:alt: BBBC013ExampleIcon
:align: left
```
@@ -100,14 +100,14 @@ U2OS citation doi: 10.1038/nmeth.2083
`````{grid}
````{grid-item}
:columns: 3
-```{image} img/example_project_icons/PLP1ExampleProjectIcon.png
+```{image} ../../img/example_project_icons/PLP1ExampleProjectIcon.png
:alt: PLP1ExampleIcon
:align: left
```
````
````{grid-item}
:columns: 9
-These human HeLa cells express either wild type or the disease-associated variant of PLP1 protein, which localizes differently than the healthy version.
+These human HeLa cells express either wild type or the disease-associated variant of PLP1 protein, which localizes differently than the healthy version.
Channel 1 is artifacts, channel 2 is fluorescently tagged protein PLP1, and channel 3 is DNA.
The human PLP1 localization data were provided by Jessica Lacoste in [Mikko Taipale's lab](http://taipalelab.org/) at University of Toronto. \
diff --git a/piximi-documentation/hyperparameters.md b/piximi-documentation/pages/technical/hyperparameters.md
similarity index 100%
rename from piximi-documentation/hyperparameters.md
rename to piximi-documentation/pages/technical/hyperparameters.md
diff --git a/piximi-documentation/technical-faq.md b/piximi-documentation/pages/technical/technical-faq.md
similarity index 75%
rename from piximi-documentation/technical-faq.md
rename to piximi-documentation/pages/technical/technical-faq.md
index 6bf34d2..1380bf9 100644
--- a/piximi-documentation/technical-faq.md
+++ b/piximi-documentation/pages/technical/technical-faq.md
@@ -10,39 +10,41 @@
- [If I run Piximi multiple times, why do I get different traiing results?](run-model-multiple-times)
(if-piximi-crashes)=
+
## If Piximi crashes, how can I recover my project?
Currently, there is no mechanism to auto-save work. It is highly recommended to manually save work periodically as you go.
The "Save project" option will save the entire state of the project, including all images and annotations made on them, and model settings (preprocessing, architecture, optimization, and dataset settings), but not including the trained model weights.
-
+
-
+
To save the trained model weights, use the "Save classifier" option.
-
+
-
+
If Piximi crashes, reload your work by using the "Open project" option to load images and project settings.
-
+
-
+
Use the "Open classifier" option to load a trained model and its parameters.
-
+
Make sure to select both the weights (model paramaters) bin file and json (model architecture) file.
-
+
-
+
(can-i-run-piximi-offline)=
+
## Can I run Piximi offline?
Yes. Once you visit the application, there is no need for an internet connection so long as you do not close or refresh the tab. If you close or refresh the tab, you will need an internet connection to reload Piximi.
@@ -54,11 +56,13 @@ Segmentation models do require an internet connection and certain segmentation m
Segmentation models that transmit data over the internet are clearly indicated.
(is-there-logging)=
+
## Is there logging?
No. Piximi does not log any information, perform any telemetry, or make any external API calls.
(what-models-are-used)=
+
## What classifier models are used?
SimpleCNN, which uses 2 convolutional layers, 2 max pooling layers, and 1 dense layer. All layers are initialized with random weights and the entire model is trained.
@@ -66,6 +70,7 @@ SimpleCNN, which uses 2 convolutional layers, 2 max pooling layers, and 1 dense
[MobileNetV1](https://github.com/tensorflow/models/blob/master/research/slim/nets/mobilenet_v1.md) (specifically, `MobileNet_v1_0.25_224`), which is a model pre-trained for image classification. Only the final convolutional layer's parameters are modified during training, the rest of the parameters in the model are frozen.
(what-if-internet-is-lost)=
+
## What if internet connection is lost while a model is training?
Once Piximi is loaded, no internet connection is necessary. You may keep working, save your project and model, and load previously saved projects and models. If you close the tab containing Piximi, or hit refresh on the browser, you will need an internet connection to reload Piximi.
@@ -73,21 +78,23 @@ Once Piximi is loaded, no internet connection is necessary. You may keep working
No internet connection is necessary to save or load projects and models.
(see-a-training-summary)=
+
## Is it possible to see a training summary?
Yes. The model summary, accuracy and loss are displayed in the Classifier dialog, and will remain there even if you leave the dialog and re-enter.
-
+
Additional metrics are available via the Evaluation dialog.
-
+
-
+
If a new model is trained however, or if the current model is re-trained, these will be lost. To avoid this, save the current model before performing any additional training.
(does-piximi-use-a-gpu)=
+
## Does Piximi use a GPU?
Yes. Piximi uses Tensorflow.js which in turn uses [WebGL](https://en.wikipedia.org/wiki/WebGL).
@@ -95,8 +102,9 @@ Yes. Piximi uses Tensorflow.js which in turn uses [WebGL](https://en.wikipedia.o
If using Chrome, users will need to enable GPU use by going into preferences -> advanced -> system, and enabling the "Use hardware acceleration when available" option.
(run-model-multiple-times)=
+
## If I run the same model multiple times, why do I get different training results?
-It is possible to get different training results when training on the same data. There are two major reasons for this: the first is a result of the random validation dataset that is selected by Piximi when you press  `Fit Classifier`. For example, the first time you fit the classifier, images 1, 2 and 3 may be selected for the validation dataset. A second, and identical, run of fit classifier might then select images 4, 5 and 6 as your validation dataset, which look different to the images selected in the first run. Validation image selection is random so that the model performance can be evaluated independently of the images selected for validation. Even if your validation data set were identical, however, your results may still end up slightly different run-to-run due to certain steps in the training process that draw on random numbers and/or shuffle the data; we do not currently but may in the future provide ways to stabilize these parameters across runs.
+It is possible to get different training results when training on the same data. There are two major reasons for this: the first is a result of the random validation dataset that is selected by Piximi when you press  `Fit Classifier`. For example, the first time you fit the classifier, images 1, 2 and 3 may be selected for the validation dataset. A second, and identical, run of fit classifier might then select images 4, 5 and 6 as your validation dataset, which look different to the images selected in the first run. Validation image selection is random so that the model performance can be evaluated independently of the images selected for validation. Even if your validation data set were identical, however, your results may still end up slightly different run-to-run due to certain steps in the training process that draw on random numbers and/or shuffle the data; we do not currently but may in the future provide ways to stabilize these parameters across runs.
Given these considerations, please do save your models frequently if you think they are performing well - you can always delete an old model later, but a generated model cannot be sure to be generated again if it hasn't been saved!
diff --git a/piximi-documentation/work-in-progress.md b/piximi-documentation/pages/technical/work-in-progress.md
similarity index 100%
rename from piximi-documentation/work-in-progress.md
rename to piximi-documentation/pages/technical/work-in-progress.md
diff --git a/piximi-documentation/classify-example-eukaryotic-image.md b/piximi-documentation/pages/tutorial/classify-example-eukaryotic-image.md
similarity index 56%
rename from piximi-documentation/classify-example-eukaryotic-image.md
rename to piximi-documentation/pages/tutorial/classify-example-eukaryotic-image.md
index 1eb2eee..0f3d85c 100644
--- a/piximi-documentation/classify-example-eukaryotic-image.md
+++ b/piximi-documentation/pages/tutorial/classify-example-eukaryotic-image.md
@@ -1,12 +1,12 @@
# Image Classification
-## 1. Load images
-To begin, we will load the images from an example dataset included in Piximi by pressing  `Open` in the top left. Select `Open` > `Project` > `Example Project` > `Human U2OS-cells example project` to get started. Alternatively, if you would like to load your own images, go to `Open` > `Image` > `New Image`.
+## 1. Load images
-The images correspond to U2OS cells co-expressing arrestin-GFP and an orphan GPCR. Upon receptor stimulation arrestin-GFP is recruited to the plasma membrane and eventually endocytosed resulting in vesicle like structures.
+To begin, we will load the images from an example dataset included in Piximi by pressing  `Open` in the top left. Select `Open` > `Project` > `Example Project` > `Human U2OS-cells example project` to get started. Alternatively, if you would like to load your own images, go to `Open` > `Image` > `New Image`.
+The images correspond to U2OS cells co-expressing arrestin-GFP and an orphan GPCR. Upon receptor stimulation arrestin-GFP is recruited to the plasma membrane and eventually endocytosed resulting in vesicle like structures.
-```{figure} ./img/user-guide-open-img.png
+```{figure} ../../img/eukaryotic-classification/load-example-light.webp
---
name: open-img
---
@@ -15,24 +15,25 @@ Open the U2OS example dataset
## 2. Categorize images
-In the `Categories` sub-menu on the left hand side you can see that there are already 3 classes defined for the U2OS example project. To turn on/off the display of images under a given label, click on the  filters icon on the right hand panel, and toggle the label of interest under `By category`. The classes are:
+In the `Categories` sub-menu on the left hand side you can see that there are already 3 classes defined for the U2OS example project. To turn on/off the display of images under a given label, click on the  filters icon on the right hand panel, and toggle the label of interest under `By category`. The classes are:
+
- Unknown
- This represents the uncategorized images. Piximi will predict which class these images belong to later
- Positive Control (vesicular GFP)
- Negative Control (cytoplasmic GFP)
-```{figure} ./img/user-guide-u2os-label-highlight.png
+```{figure} ../../img/eukaryotic-classification/human-u20s-category-light.webp
---
name: u2os-labels
---
-Explore the category menu. Turn on/off (/) a particular category to show/hide only those images.
+Explore the category menu. Turn on/off (/) a particular category to show/hide only those images.
```
-
-Ensure no images are currently selected by clicking the  `Deselect` icon. Then, single-click to select 2-3 images from the unknown category that best fit the `Negative Control` category. Once selected, click `Categorize` in the top right and select `Negative Control`. Do the same for 2-3 `Positive Control` images.
+Ensure no images are currently selected by clicking the  `Deselect` icon. Then, single-click to select 2-3 images from the unknown category that best fit the `Negative Control` category. Once selected, click `Categorize` in the top right and select `Negative Control`. Do the same for 2-3 `Positive Control` images.
+
```{math}
:label: accuracy_equation
Accuracy = \frac{\text{Number of correct predictions}}{\text{Total number of predictions}}
```
-`Validation Accuracy` is the accuracy when the model examines the **validation** subset of the data.
+`Validation Accuracy` is the accuracy when the model examines the **validation** subset of the data.
```{admonition} Validation accuracy vs accuracy
:class: tip, dropdown
@@ -93,11 +95,11 @@ If you notice that your `Validation accuracy` value decreases as epochs increase
This is a result of **overfitting** as your model begins to pick up features within your image, such as noise, that are not relevant to classification. In essence, overfitting is when the model memorizes the answer to a specific question, rather than determining the answer from scratch itself.
```
-Loss is another metric that is calculated on the training and validation subsets of data and are depicted as loss and validation loss, respectively. Loss represents a summation of the errors the model has made during classification.
+Loss is another metric that is calculated on the training and validation subsets of data and are depicted as loss and validation loss, respectively. Loss represents a summation of the errors the model has made during classification.
-You can now exit the `Fit Model` settings by clicking the  in the top left of the dialog.
+You can now exit the `Fit Model` settings by clicking the  in the top left of the dialog.
-```{figure} ./img/user-guide-exit-fit.png
+```{figure} ../../img/eukaryotic-classification/human-u20s-fit-dialog-exit-light.webp
---
name: fit-exit
---
@@ -114,21 +116,20 @@ Piximi does not currently have a hold-out test-like set.
## 4. Predict classes for unlabelled data
-Once your model has been trained you can click  `Evaluate` to see in-depth metrics on how well the model performed. You can then click  `Predict` to run the trained model on the unannotated data. Once an image has been classified you will see the  color on the image thumbnail update to that particular class. At this stage, you may inspect the predicted classes and either accept the predictions by clicking and holding  `Accept Predictions`or reject them by clicking  `Clear Predictions`. Depending on the performance of the model, annotating further images based on the predictions and/or adjusting the `Fit Model` settings may be desired.
+Once your model has been trained you can click  `Evaluate` to see in-depth metrics on how well the model performed. You can then click  `Predict` to run the trained model on the unannotated data. Once an image has been classified you will see the  color on the image thumbnail update to that particular class. At this stage, you may inspect the predicted classes and either accept the predictions by clicking and holding  `Accept Predictions`or reject them by clicking  `Clear Predictions`. Depending on the performance of the model, annotating further images based on the predictions and/or adjusting the `Fit Model` settings may be desired.
-```{figure} ./img/user-guide-u2os-run-predict.png
+```{figure} ../../img/eukaryotic-classification/human-u20s-predict-light.webp
---
name: run-predict
---
Predict the class of unknown images using your trained model.
```
-
+
+````-->
```{admonition} Copyright
:class: seealso
The [BBBC016](https://bbbc.broadinstitute.org/BBBC016) images used here are licensed under a [Creative Commons Attribution 3.0 Unported License](https://creativecommons.org/licenses/by/3.0/) by Ilya Ravkin.
```
-
-
diff --git a/piximi-documentation/measurements.md b/piximi-documentation/pages/tutorial/creating-measurements.md
similarity index 75%
rename from piximi-documentation/measurements.md
rename to piximi-documentation/pages/tutorial/creating-measurements.md
index 54f1cdd..d72f2c0 100644
--- a/piximi-documentation/measurements.md
+++ b/piximi-documentation/pages/tutorial/creating-measurements.md
@@ -1,21 +1,21 @@
-# Measurements
+# Creating Measurements
-While sometimes, the goal of an experiment is just to find objects or to classify images and/or objects, other times, we want to go deeper and learn more about the images and/or objects we care about. This is where *Measurements* come in.
+While sometimes, the goal of an experiment is just to find objects or to classify images and/or objects, other times, we want to go deeper and learn more about the images and/or objects we care about. This is where _Measurements_ come in.
Right now, Piximi supports two broad classes of measurements - intensity measurements, which work for images OR objects, and shape/geometry measurements, which work for objects alone. More categories of measurements will be rolling out in time - stay tuned!
-Measurements are available from the  icon in the top bar of Piximi.
+Measurements are available from the  icon in the top bar of Piximi.
-Measurements are created in individual tables - each table is specific to one "kind" - either images or a type of segmented object (such as cells). You can create as many tables as you like.
-Each table has measurements broken down by *Splits* - essentially, what are the different conditions in this experiment (if any) and how should we parse them? Each table has its own set of splits, which you can adjust at any time.
+Measurements are created in individual tables - each table is specific to one "kind" - either images or a type of segmented object (such as cells). You can create as many tables as you like.
+Each table has measurements broken down by _Splits_ - essentially, what are the different conditions in this experiment (if any) and how should we parse them? Each table has its own set of splits, which you can adjust at any time.
-# Creating Measurements in Piximi
+# Create Measurements in Piximi
-1. Click the [Measurements](./icons/ruler-icon.svg) icon in the top bar of Piximi.
+1. Click the [Measurements](../../icons/ruler-icon.svg) icon in the top bar of Piximi.
2. Click the `+` button next to the word `Tables` to create a new measurement table. You can make as many tables as you like, to express different images or objects, broken down by different subsets.
-```{figure} ./img/measurement-walkthrough/measure-00-pre-measurement.png
+```{figure} ../../img/measurement-walkthrough/measure-00-pre-measurement.png
---
name: measurement-00-table-plus
scale: 50%
@@ -25,7 +25,7 @@ The `+` button at the top of the Tables menu allows you to add a new table
3. Select a `kind` you'd like to make measurements of. Reminder, `kind`s include the images, as well as any objects you created via segmentation or annotation.
-```{figure} ./img/measurement-walkthrough/measure-01-create-table.png
+```{figure} ../../img/measurement-walkthrough/measure-01-create-table.png
---
name: measure-01-create-table
scale: 50%
@@ -35,7 +35,7 @@ Select what kind of measurement table you'd like to make
4. Piximi will begin to calculate metrics in the background. Depending on the number of examples present in that `kind`, this may take a few minutes - an indicator will show you the progress made.
-```{figure} ./img/measurement-walkthrough/measure-02-spin-while-creating-table.png
+```{figure} ../../img/measurement-walkthrough/measure-02-spin-while-creating-table.png
---
name: measure-02-spin-while-creating-table
scale: 50%
@@ -45,7 +45,7 @@ The circular indicator next to `Tables` indicates how far through measurement ge
5. Once the initial pre-calculations are done, you can select which measurements you'd like to make, as well as how you'd like to subset "split" the data (currently by `Category` or by `Partition`; if you only have one category, tha's fine, but you do have to click something here!). As before, this will take some time, so just leave the Piximi tab open and the circular indicator will let you know how long this is going to take.
-```{figure} ./img/measurement-walkthrough/measure-03-select-splits-and-measurements.png
+```{figure} ../../img/measurement-walkthrough/measure-03-select-splits-and-measurements.png
---
name: measure-03-select-splits-and-measurements
scale: 50%
@@ -55,7 +55,7 @@ Select the split or splits you want to measure for each table, as well as the me
6. Once your measurements have been generated, they will appear in the main area in a data grid! In the grid, you can see the measurement name, the subset (or `Split`) of the data represented, and the Mean, Median, and Standard deviation of the data for that feature in that subset.
-```{figure} ./img/measurement-walkthrough/measure-04-data-grid.png
+```{figure} ../../img/measurement-walkthrough/measure-04-data-grid.png
---
name: measure-04-data-grid
---
@@ -64,7 +64,7 @@ The Piximi data grid
7. If you'd like to plot data in Piximi, simply navigate to the data grid containing the data you want to plot and hit the Plots tab at the top. You can make as many plots as you would like by hitting the `+` button on the plots tab. Plots can also be exported by saving to PNG. Piximi includes several plot types, including scatter plots, swarm plots, and histograms.
-```{figure} ./img/measurement-walkthrough/measure-05-scatter.png
+```{figure} ../../img/measurement-walkthrough/measure-05-scatter.png
---
name: measure-05-scatter
scale: 50%
@@ -72,7 +72,7 @@ scale: 50%
Scatter plots generated in Piximi can have X, Y, and Size measurements selected; they can also show colors according to the selected splits (with many colormaps to choose from). Note the "Save to PNG" button in the bottom right!
```
-```{figure} ./img/measurement-walkthrough/measure-06a-swarm-no-category.png
+```{figure} ../../img/measurement-walkthrough/measure-06a-swarm-no-category.png
---
name: measure-06a-swarms
scale: 50%
@@ -80,7 +80,7 @@ scale: 50%
Swarm plots can be shown on their own...
```
-```{figure} ./img/measurement-walkthrough/measure-06b-swarm-with-category.png
+```{figure} ../../img/measurement-walkthrough/measure-06b-swarm-with-category.png
---
name: measure-06b-swarms
scale: 50%
@@ -88,7 +88,7 @@ scale: 50%
... or with a summary boxplot overlaid.
```
-```{figure} ./img/measurement-walkthrough/measure-07-histogram.png
+```{figure} ../../img/measurement-walkthrough/measure-07-histogram.png
---
name: measure-07-histogram
scale: 50%
@@ -96,9 +96,9 @@ scale: 50%
You can choose to represent your data as a histogram, with a custom number of bins.
```
-If you want to do further analyses of your data (and you should)!, you can export your data tables from the measurements menu - this will let you do whatever downstream analysis you like.
+If you want to do further analyses of your data (and you should)!, you can export your data tables from the measurements menu - this will let you do whatever downstream analysis you like.
-```{figure} ./img/measurement-walkthrough/measure-08-export-measurements.png
+```{figure} ../../img/measurement-walkthrough/measure-08-export-measurements.png
---
name: measure-08-export-measurements
scale: 50%
diff --git a/piximi-documentation/segmentation.md b/piximi-documentation/pages/tutorial/segmentation-tutorial.md
similarity index 52%
rename from piximi-documentation/segmentation.md
rename to piximi-documentation/pages/tutorial/segmentation-tutorial.md
index 451b99c..f6c2650 100644
--- a/piximi-documentation/segmentation.md
+++ b/piximi-documentation/pages/tutorial/segmentation-tutorial.md
@@ -1,15 +1,13 @@
-# Segmentation
-
# Image Segmentation
The image segmentation module allows researchers to quickly identify cells or nuclei by selecting pre-trained segmentation models. Users can choose a model from the available options and apply it to images that are opened and selected in Piximi for inference.
## 1. Load images
-To begin, loading the images from an example dataset included in Piximi by pressing  `Open` in the top left. Select `Image` > `example image` > `U2OS cell-painting experiment ` to get started. Alternatively, if you would like to load your own images, press  `Open image` in the top right.
+To begin, loading the images from an example dataset included in Piximi by pressing  `Open` in the top left. Select `Image` > `example image` > `U2OS cell-painting experiment ` to get started. Alternatively, if you would like to load your own images, press  `Open image` in the top right.
The images correspond to U2OS cells U2OS cells treated with an RNAi reagent
-(https://portals.broadinstitute.org/gpp/public/clone/details?cloneld=TRCN0000195467) and stained for a cell-painting experiment.
+(https://portals.broadinstitute.org/gpp/public/clone/details?cloneld=TRCN0000195467) and stained for a cell-painting experiment.
## 2. Load Models
@@ -17,55 +15,65 @@ Currently, the annotator provides five pre-trained models, each designed for spe
- **Cellpose**: A generalist algorithm for cellular segmentation and trained on fluorecence images
- **Stardistfluo**: Trained on fluorecence images, ideal for identifying objects with star-convex shapes
-- **StardistVHE**: To identify nuclei in hematoxylin and eosin (H&E) stained image.
+- **StardistVHE**: To identify nuclei in hematoxylin and eosin (H&E) stained image.
- StardistFluo
- **COCO-SSD**: To identify objects in “natural images” (or photographs) of 80 different classes (such as humans and kites) using the COCO format
- **GlandSegmentation**: To segment intestinal glands trained on the Gland Segmentation in Colon Histology Images Challenge Contest (GlaS)29
In the 'Learning task' sub-menu on the left-hand side, click the 'Segmentation' button to switch the classification to cell segmentation.
-
-
+
+
-*Segmentation interface*
+_Segmentation interface_
-
+
+
-*Segmentation model selection*
+_Segmentation model selection_
+
+
+
+
+Note: Cellpose is currently unique in that it runs on the AI4Life project’s BioEngine30 server while StarDist, like other Piximi models, runs client-only in the user’s own browser without data leaving their machine.
Stay tuned for 'upload local' and fetch 'remote feaure'!
## 3. Run the model
-click on  'predict model' to run the model you selected for the image selected for segmentation
+
+click on  'predict model' to run the model you selected for the image selected for segmentation
-
-*Model prediction*
+
+
-
-
-*Examples of segmentation output*
+
+
+
+_Examples of segmentation output_
+
+
+
+_Schematic representation of FOXO1A mechanism_
+
+
.
+
+
+
+
+
+
+_Loading a project file._
+
+
+
+
+
+
+_Viewing the project images._
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+_Viewing the project images in the Image Viewer._
+
+
+
+
+
+
+_Viewing the cellpose_cells kind._
+
+
+
+
+
+
+_Action Drawer Classifier Section._
+
+
+
+
+
+
+_Create Category Button._
+
+
+
+
+
+
+_Classifying individual cells based on GFP presence and localization._
+
+
- Fit Model” icon to open the model hyperparameter settings. For today’s exercise, we’ll adjust a few parameters:
+- Click on “Architecture Settings” and set the Model Architecture to **SimpleCNN**.
+- Update the Input Dimensions to:
+
+ - Input rows: 48
+ - Input cols: 48
+ - Channels: 3 (since our images are in RGB format)
+
+ (You can change to other numbers such as 64, 128)
+
+- In the “Data Partitioning” section, set the Training Percentage to 0.75, which reserves 25% of the labeled data for validation.
+
+
+
+
+
+
+_Classifier Model Setup._
+
+
+
+
+
+
+_Training History Plots._
+
+
+
+
+
+
+_Training Run Evaluation._
+
+
+
+
+
+
+_Predict Classifier._
+
+
+
+
+
+
+_Accept Predictions._
+
+
+
+
+
+
+_Navigate to Measurements._
+
+
+
+
+
+
+_Create "**Image**" measurement table._
+
+
+
+
+
+
+_Calculated Measurements._
+
+
+
+
+
+
+_Measurement Plots._
+
+
+
+
+
+
+_Calculated Measurements._
+
+
.
+
+##### 11. **Supporting Information**
+
+Check out the Piximi paper: [https://www.biorxiv.org/content/10.1101/2024.06.03.597232v2](https://www.biorxiv.org/content/10.1101/2024.06.03.597232v2)
+
+Check out the Piximi documentation:[Piximi documentation](https://documentation.piximi.app/intro.html):[https://documentation.piximi.app/intro.html](https://documentation.piximi.app/intro.html)
+
+Report bugs/errors or request features [https://github.com/piximi/documentation/issues](https://github.com/piximi/documentation/issues)
diff --git a/piximi-documentation/pages/tutorial/translocation_tutorial_ES.md b/piximi-documentation/pages/tutorial/translocation_tutorial_ES.md
new file mode 100644
index 0000000..135cc6b
--- /dev/null
+++ b/piximi-documentation/pages/tutorial/translocation_tutorial_ES.md
@@ -0,0 +1,379 @@
+# Tutorial Inicial de Piximi (ESPAÑOL)
+
+> **Segmentación y clasificación sin instalación en el navegador**
+>
+> Beth Cimini, Le Liu, Esteban Miglietta, Paula Llanos, Nodar Gogoberidze
+>
+> Instituto Broad del MIT y Harvard, Cambridge, MA.
+
+### **Información general:**
+
+#### **¿Qué es Piximi?**
+
+Piximi es una herramienta moderna de análisis de imágenes tomando ventaja de varios métodos de _deep learning_, sin requerir conocimientos de programación. Implementado como una aplicación web en [https://piximi.app/](https://piximi.app/), Piximi no requiere instalación y se puede acceder desde cualquier navegador web moderno. Su arquitectura de cliente único preserva la seguridad de los datos del investigador ejecutando todos los cálculos localmente\*.
+
+Piximi es interoperable con herramientas y flujos de trabajo existentes, ya que admite la importación y exportación formatos de datos y modelos comunes. La interfaz intuitiva y el fácil acceso a Piximi permiten a los biólogos obtener información sobre las imágenes en tan sólo unos minutos. Piximi tiene como objetivo llevar el análisis de imágenes basado en _deep learning_ a una comunidad más amplia mediante la eliminación de las barreras de entrada.
+
+\* excepto las segmentaciones mediante Cellpose, que se envían a un servidor remoto (con el permiso del usuario).
+
+Funciones básicas: **Anotador, Segmentador, Clasificador, Mediciones.**
+
+#### **Objetivo del ejercicio**
+
+En este ejercicio, se familiarizará con las principales funcionalidades de Piximi de anotación, segmentación, clasificación, medición y visualización y lo utilizará para analizar un conjunto de imágenes de muestra de un experimento de translocación. El objetivo de este experimento es determinar la **dosis efectiva más baja** de Wortmannin requerida para inducir la localización nuclear de FOXO1A etiquetada con GFP (Figura 31). Segmentará las imágenes utilizando uno de los modelos de _deep learning_ disponibles en Piximi. Comprobará y curará la segmentación y luego entrenará un clasificador de imágenes para clasificar las células individuales como teniendo «GFP nuclear», «GFP citoplasmática» o «sin GFP». Por último, realizará mediciones y las representará gráficamente para responder a la pregunta biológica.
+
+#### **Contexto del experimento de muestra**
+
+En este experimento, los investigadores tomaron imágenes de células U2OS de osteosarcoma (cáncer de hueso) fijadas que expresaban una proteína de fusión FOXO1A-GFP y tiñeron con DAPI para marcar los núcleos. FOXO1 es un factor de transcripción que desempeña un papel clave en la regulación de la gluconeogénesis y la glicogenólisis a través de la señalización de insulina. FOXO1A se desplaza dinámicamente entre el citoplasma y el núcleo en respuesta a diversos estímulos. Wortmannin, un inhibidor de PI3K, puede bloquear la exportación nuclear, lo que resulta en la acumulación de FOXO1A en el núcleo.
+
+
+
+
+
+_Representación esquemática del mecanismo de acción de FOXO1A._
+
+
+
+
+
+
+_Cargando el archivo de proyecto._
+
+
+
+
+
+
+_Visualización de las imágenes del proyecto._
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+_Visualización de las imágenes del proyecto en el Visor de imágenes._
+
+
+
+
+
+
+_Visualización del tipo «cellpose_cells»._
+
+
+
+
+
+
+_Sección de clasificadores de Action Drawer._
+
+
+
+
+
+
+_Botón Crear categoría._
+
+
+
+
+
+
+_Clasificando células individuales en base a la presencia de GFP y su localización._
+
+
+
+
+
+
+_Configuración del modelo clasificador._
+
+
+
+
+
+
+_Gráficos del historial de entrenamiento._
+
+
+
+
+
+
+_Evaluación de la ejecución de entrenamiento._
+
+
+
+
+
+
+_Predecir clasificador._
+
+
+
+
+
+
+_Aceptar predicciones._
+
+
+
+
+
+
+_Navegar a Medidas._
+
+
+
+
+
+
+_Crear tabla de medidas "**Image**"._
+
+
+
+
+
+
+_Medidas calculadas._
+
+
+
+
+
+
+_Parcelas de medición._
+
+
+
+
+
+
+_Graficar los resultados._
+
+
+
+
+
+_Schematic representation of FOXO1A mechanism_
+
+
+
+
+
+
+_Carregando um arquivo de projeto._
+
+
+
+
+
+
+_Explorando as imagens e rótulos._
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+_Visualizando as imagens do projeto no Visualizador de Imagens._
+
+
+
+
+
+
+_Visualizando o tipo "cellpose_cells"._
+
+
+
+
+
+
+_Seção Classificador do Action Drawer._
+
+
+
+
+
+
+_Botão Criar Categoria._
+
+
+
+
+
+
+_Classificação de células individuais com base na presença e localização de GFP._
+
+
+
+
+
+
+_Configuração do modelo classificador._
+
+
+
+
+
+
+_Gráficos do histórico de treinamento._
+
+
+
+
+
+
+_Avaliação da execução de treinamento._
+
+
+
+
+
+
+_Prever classificador._
+
+
+
+
+
+
+_Aceitar previsões._
+
+
+
+
+
+
+_Navegar para Medidas._
+
+
+
+
+
+
+_Criar tabela de medidas "**Image**"._
+
+
+
+
+
+
+_Medidas calculadas._
+
+
+
+
+
+
+_Gráficos de medição._
+
+
+
+
+
+
+_Resultados do gráfico._
+
+
.
-
-```{figure} ./img/tutorial_images/Figure2.png
-:width: 600
-:align: center
-
-Loading a project file.
-```
-
-##### 2. **Check the loaded images and explore the Piximi interface**
-
-These 17 images represent Wortmannin treatments at eight different concentrations (expressed in nM), as well as mock treatments (0nM). Note the DAPI channel (Nuclei) is shown in magenta and that the GFP channel (FOXOA1) is shown in green.
-
-As you hover over the image, color labels are displayed on the left corner of the images. These annotations are from metadata in the zipped file we just uploaded. In this tutorial, the different colored labels indicate the concentration of Wortmannin, while the numbers represent the number of images in each category.
-
-Optionally, you can annotate the images manually by clicking “+ Category”, entering your label, and then selecting the image by clicking the images annotating the selected images by clicking **“Categorize”**. In this tutorial, we’ll skip this step since the labels were already uploaded at the beginning.
-
-```{figure} ./img/tutorial_images/Figure3.png
-:width: 600
-:align: center
-
-Exploring the images and labels.
-```
-
-##### 3. **Segment Cells - find out the cells from the background**
-
-
- 🔴 TO DO
-
-* To start the prediction on all images, click “Select All Images” in the top panel as shown in Figure 23.
-* Change the Learning Task to “SEGMENTATION” (Figure 24, Arrow 1).
-
-* Click on “+ LOAD MODEL” (Arrow 2) and the window will pop up, allowing you to choose a pretrained model (Arrow 3). For today’s exercise, select “Cellpose” (Arrow 4). More information about the model supported can be found [here](https://documentation.piximi.app/segmentation.html).
-* Click “Open Segmentation Model” (Arrow 5) to load your model and select it. Finally, click “Predict Model” (Arrow 5). You’ll see the prediction progress displayed in the top left corner beneath the Piximi logo 
.
-* It will take a few minutes to finish the segmentation.
-
-
-```{figure} ./img/tutorial_images/Figure4.png
-:width: 600
-:align: center
-
-Loading a segmentation model.
-```
-
-Please note that the previous steps were performed on your local machine, meaning your images are stored locally. However, Cellpose inference runs in the cloud, which means your images will be uploaded for processing. If your images are highly sensitive, please exercise caution when using cloud-based services.
-
-##### 4. **Visualize segmentation result and fix the segmentation errors**
-
-🔴 TO DO
-
-* Click on the **CELLPOSE_CELLS** tab to check the individual cells that have been segmented. Click on the “IMAGE” tab and then “Annotate”, you can check the segmentation on the whole image.
-
-```{figure} ./img/tutorial_images/Figure5.png
-:width: 600
-:align: center
-
-Piximi's annotator tool.
-```
-
-* Optionally, here you can manually refine the segmentation using the annotator tools. The Piximi annotator provides several options to **add**, **subtract**, or **intersect** annotations. Additionally, the **selection tool** allows you to **resize** or **delete** specific annotations. To begin editing, select specific or all images by clicking the checkbox at the top.
-* Optionally, you can adjust channels: Although there are two channels in this experiment, the nuclei signal is duplicated in both the red and green channels. This design is intended to be **color-blind friendly** and to produce a **magenta color** for nuclei. The **green channel** also includes cytoplasmic signals.
-
-Another reason for duplicating the channels is that some models—such as the **Cellpose model** we used today—require a **three-channel** input.
-
-* You can choose to manually segment the cells to generate masks for ground truth data.
-
-##### **Classify cells**
-
-Reason for doing this: We want to classify the 'CELLPOSE\_CELLS' based on GFP distribution (on Nuclei, cytoplasm, or no GFP) without manually labeling all of them. To do this, we can use the classification function in Piximi, which allows us to train a classifier using a small subset of labeled data and then automatically classify the remaining cells.
-
- 🔴 TO DO
-
-* Go to the **CELLPOSE_CELLS** tab that displays the segmented objects (arrow 1, figure 26)
-* Click on the **Classification** tab on the left panel (arrow 2, figure 26).
-* Create new categories by clicking **“+ Category”**. Adding “Cytoplasmatic_GFP”, “Nuclear_GFP”, “No_GFP” three categories (Arrow 3, Figure 26).
-* Click on the images that match your criteria. You can select multiple cells by holding **Command (⌘)** on Mac or **Shift** on Linux. Aim to assign **\~20–40 cells per category**. Once selected, click **“Categorize”** to assign the labels to the selected cells.
-
-```{figure} ./img/tutorial_images/Figure6.png
-:width: 600
-:align: center
-
-Classifying individual cells based on GFP presence and localization.
-```
-
-##### 6. **Train the Classifier model**
-
- 🔴 TO DO
-
-* Click the ”
- Fit Model” icon to open the model hyperparameter settings. For today’s exercise, we’ll adjust a few parameters:
-* Click on “Architecture Settings” and set the Model Architecture to **SimpleCNN**.
-* Update the Input Dimensions to:
- - Input rows: 48
- - Input cols: 48
- - Channels: 3 (since our images are in RGB format)
-
- (You can change to other numbers such as 64, 128)
-
-```{figure} ./img/tutorial_images/Figure7.png
-:width: 600
-:align: center
-
-Classifier model setup.
-```
-
-* Click on the “Dataset Setting” tab and set the Training Percentage to 0.75, which reserves 25% of the labeled data for validation.
-* When you click **"Fit Classifier"** in Piximi, two training plots will appear “**Accuracy vs Epochs”** and **“Loss vs Epochs”**. Each plot shows curves for both **training** and **validation** data.
-* In the **accuracy plot**, you’ll see how well the model is learning. Ideally, both training and validation accuracy should increase and stay close.
-* In the **loss plot**, lower values mean better performance. If validation loss starts rising while training loss keeps dropping. g, the model might be overfitting.
-
-These plots help you understand how the model is learning and whether adjustments are needed.
-
-##### 7. **Evaluate model:**
-
- 🔴 TO DO
-
-```{figure} ./img/tutorial_images/Figure8.png
-:width: 400
-:align: center
-
-Classifier training and validation.
-```
-
-* Click **“Predict Model” (figure 28, arrow 1)** to apply the model we just trained. This step will generate predictions on the cells we did not annotate.
-* You can review the predictions in the CELLPOSE_CELLS tab and delete any wrongly assigned categories.
-* Optionally, you can continue using the labels to refine the ground truth and improve the classifier. This process is part of the **Human-in-the-loop classification**, where you iteratively correct and train the model based on human input.
-* Click **“Evaluate Model” (figure 28, arrow 2)** to evaluate the model we just trained. The confusion metrics and evaluation metrics can be compared to the ground truth.
-* Click "Accept Prediction (Hold)”, to assign the predicted labels to all the objects.
-
-##### 8. **Measurement**
-
-Once you are satisfied with the classification, we will proceed to measure the objects. The goal of today’s exercise is to determine the minimum concentration of Wortmannin required to block the export of FOXO1A-GFP from the nuclei. To do this, we can measure the total GFP intensity at either the image level or the object level.
-
- 🔴 TO DO
-
-* Click “Measurement” in the top right corner.
-* Click Tables (Arrow 1) and select Image and click “Confirm” (Arrow 2).
-* Choose "MEASUREMENT" in the left panel, note the measurement step may take some time to process.
-* Click on 'Category' to include all categories in the measurement.
-* "Under 'Total', click on 'Channel 1' (Arrow 3) to select the measurement for GFP. You will see the measurement in the “DATA GRID” tab. Measurements are presented as either mean or median values, and the full dataset is available upon exporting the .csv file.
-
-```{figure} ./img/tutorial_images/Figure9.png
-:width: 600
-:align: center
-
-Add measurements.
-```
-
-##### 9. **Visualization**
-
-After generating the measurements, you can plot the measurements.
-
- 🔴 TO DO
-
-* Click on 'PLOTS' (Figure 30, Arrow 1) to visualize the measurements.
-* Set the plot type to 'Swarm' and choose a color theme based on your preference.
-* Select 'Y-axis' as 'intensity-total-channel-1' and set 'SwarmGroup' to 'category'; this will generate a curve showing how GFP intensity varies across different categories (Figure 30, Arrow 2).
-* Selecting 'Show Statistics' will display the mean, as well as the upper and lower confidence boundaries, on the plot.
-* Optionally, you can experiment with different plot types and axes to see if the data reveals additional insights.
-
-```{figure} ./img/tutorial_images/Figure10.png
-:width: 600
-:align: center
-
-Plot results.
-```
-
-##### 10. **Export results and save the project**
-
- 🔴 TO DO
-
-* Click “SAVE” in the top left corner to save the entire project. You'll see the Piximi logo animation as the save progresses .
-
-##### 11. **Supporting Information**
-
-Check out the Piximi paper: [https://www.biorxiv.org/content/10.1101/2024.06.03.597232v2](https://www.biorxiv.org/content/10.1101/2024.06.03.597232v2)
-
-Check out the Piximi documentation:[Piximi documentation](https://documentation.piximi.app/intro.html):[https://documentation.piximi.app/intro.html](https://documentation.piximi.app/intro.html)
-
-Report bugs/errors or request features [https://github.com/piximi/documentation/issues](https://github.com/piximi/documentation/issues)
diff --git a/piximi-documentation/translocation_tutorial_ES.md b/piximi-documentation/translocation_tutorial_ES.md
deleted file mode 100644
index a44baf9..0000000
--- a/piximi-documentation/translocation_tutorial_ES.md
+++ /dev/null
@@ -1,227 +0,0 @@
-# Tutorial inicial de Piximi (ESPAÑOL)
-
-## Segmentación y clasificación sin instalación en el navegador
-
-Beth Cimini, Le Liu, Esteban Miglietta, Paula Llanos, Nodar Gogoberidze
-
-Instituto Broad del MIT y Harvard, Cambridge, MA.
-
-### **Información general:**
-
-#### **¿Qué es Piximi?**
-
-Piximi es una herramienta moderna de análisis de imágenes tomando ventaja de varios métodos de *deep learning*, sin requerir conocimientos de programación. Implementado como una aplicación web en [https://piximi.app/](https://piximi.app/), Piximi no requiere instalación y se puede acceder desde cualquier navegador web moderno. Su arquitectura de cliente único preserva la seguridad de los datos del investigador ejecutando todos los cálculos localmente\*.
-
-Piximi es interoperable con herramientas y flujos de trabajo existentes, ya que admite la importación y exportación formatos de datos y modelos comunes. La interfaz intuitiva y el fácil acceso a Piximi permiten a los biólogos obtener información sobre las imágenes en tan sólo unos minutos. Piximi tiene como objetivo llevar el análisis de imágenes basado en *deep learning* a una comunidad más amplia mediante la eliminación de las barreras de entrada.
-
-\* excepto las segmentaciones mediante Cellpose, que se envían a un servidor remoto (con el permiso del usuario).
-
-Funciones básicas: **Anotador, Segmentador, Clasificador, Mediciones.**
-
-#### **Objetivo del ejercicio**
-
-En este ejercicio, se familiarizará con las principales funcionalidades de Piximi de anotación, segmentación, clasificación, medición y visualización y lo utilizará para analizar un conjunto de imágenes de muestra de un experimento de translocación. El objetivo de este experimento es determinar la **dosis efectiva más baja** de Wortmannin requerida para inducir la localización nuclear de FOXO1A etiquetada con GFP (Figura 31). Segmentará las imágenes utilizando uno de los modelos de *deep learning* disponibles en Piximi. Comprobará y curará la segmentación y luego entrenará un clasificador de imágenes para clasificar las células individuales como teniendo «GFP nuclear», «GFP citoplasmática» o «sin GFP». Por último, realizará mediciones y las representará gráficamente para responder a la pregunta biológica.
-
-#### **Contexto del experimento de muestra**
-
-En este experimento, los investigadores tomaron imágenes de células U2OS de osteosarcoma (cáncer de hueso) fijadas que expresaban una proteína de fusión FOXO1A-GFP y tiñeron con DAPI para marcar los núcleos. FOXO1 es un factor de transcripción que desempeña un papel clave en la regulación de la gluconeogénesis y la glicogenólisis a través de la señalización de insulina. FOXO1A se desplaza dinámicamente entre el citoplasma y el núcleo en respuesta a diversos estímulos. Wortmannin, un inhibidor de PI3K, puede bloquear la exportación nuclear, lo que resulta en la acumulación de FOXO1A en el núcleo.
-
-```{figure} ./img/tutorial_images/Figure1.png
-:width: 300
-:align: center
-
-Representación esquemática del mecanismo de acción de FOXO1A.
-```
-
-#### **Materiales necesarios para este ejercicio**
-
-Los materiales necesarios para este ejercicio pueden descargarse de: [PiximiTutorial](./downloads/Piximi_Translocation_Tutorial_RGB.zip). El archivo «Piximi Translocation Tutorial RGB.zip» contiene un proyecto de Piximi que incluye todas las imágenes, ya etiquetadas con el tratamiento correspondiente (concentración de Wortmannin o Control). ¡Descargue este archivo pero **NO lo descomprima**!
-
-#### **Instrucciones para el ejercicio**
-
-Lea los pasos que se indican a continuación y siga las instrucciones donde se indican. Los pasos en los que debe averiguar una solución están marcados con 🔴 PARA HACER.
-
-##### 1. **Cargar el proyecto Piximi**
-
-🔴 PARA HACER
-
-* Inicia Piximi en:[https://piximi.app/](https://piximi.app/)
-
-* Cargar el proyecto de ejemplo: Haga clic en «Abrir» \- “Proyecto” \- «Proyecto desde Zip», como se muestra en la figura 32 para cargar un archivo de proyecto para este tutorial desde Zip, y opcionalmente puede cambiar el nombre del proyecto en el panel superior izquierdo, como «Ejercicio Piximi». A medida que se carga, se puede ver la progresión en la esquina superior izquierda logotipo .
-
-
-```{figure} ./img/tutorial_images/Figure2.png
-:width: 600
-:align: center
-
-Cargando el archivo de proyecto.
-```
-
-##### 2. **Compruebe las imágenes cargadas y explore la interfaz Piximi**
-
-Estas 17 imágenes representan tratamientos con Wortmannin a ocho concentraciones diferentes (expresadas en nM), así como tratamientos con sólo vehículo (0nM). Observe que el canal DAPI (Núcleos) se muestra en magenta y que el canal GFP (FOXOA1) se muestra en verde.
-
-Al pasar el cursor por encima de la imagen, aparecen etiquetas de color en la esquina izquierda de las imágenes. Estas anotaciones proceden de los metadatos del archivo comprimido que acabamos de cargar. En este tutorial, las etiquetas de diferentes colores indican la concentración de Wortmannin, mientras que los números representan el número de imágenes en cada categoría.
-
-Opcionalmente, puede anotar las imágenes manualmente haciendo clic en «+ Category», introduciendo su etiqueta, y luego seleccionando la imagen haciendo clic en las imágenes anotando las imágenes seleccionadas haciendo clic en **«Categorize»**. En este tutorial, nos saltaremos este paso ya que las etiquetas ya estaban cargadas al principio.
-
-```{figure} ./img/tutorial_images/Figure3.png
-:width: 600
-:align: center
-
-Explorando las imágenes y etiquetas.
-```
-
-##### 3. **Segmentar Células - diferenciar las células del *background***.
-
- 🔴 PARA HACER
-
-* Para iniciar la predicción en todas las imágenes, haga clic en «Seleccionar todas las imágenes» en el panel superior, como se muestra en la Figura 33.
-* Cambie la Tarea de Aprendizaje a «SEGMENTATION» (Figura 34, Flecha 1).
-* Haga clic en «+ LOAD MODEL» (Flecha 2) y aparecerá una ventana que le permitirá elegir un modelo pre-entrenado (Flecha 3). Para el ejercicio de hoy, seleccione «Cellpose» (Flecha 4). Puede encontrar más información sobre el modelo admitido [aquí](https://documentation.piximi.app/segmentation.html).
-* Haga clic en «Open Segmentation Modeln» (Flecha 5\) para cargar su modelo y seleccionarlo. Por último, haga clic en «Predict model» (Flecha 5). Verá el progreso de la predicción en la esquina superior izquierda debajo del logo de Piximi .
-* Tardará unos minutos en finalizar la segmentación.
-
-
-```{figure} ./img/tutorial_images/Figure4.png
-:width: 600
-:align: center
-
-Cargando un modelo de segmentación.
-```
-
-Tenga en cuenta que los pasos anteriores se realizaron en su computadora local, lo que significa que sus imágenes se almacenan localmente. Sin embargo, la inferencia de Cellpose se ejecuta en la nube, lo que significa que sus imágenes se cargarán para su procesamiento. Si sus imágenes son altamente sensibles, por favor tenga cuidado cuando utilice servicios basados en la nube.
-
-##### 4. **Visualice el resultado de la segmentación y corrija los errores de segmentación**
-
- 🔴 PARA HACER
-
-* Haga clic en la pestaña **CELLPOSE_CELLS** para comprobar las células individuales que se han segmentado. Haga clic en la pestaña «IMAGE» y luego en «Annotate», puede comprobar la segmentación en toda la imagen.
-
-```{figure} ./img/tutorial_images/Figure5.png
-:width: 600
-:align: center
-
-La herramienta de anotación de Piximi.
-```
-
-* Opcionalmente, aquí puede refinar manualmente la segmentación utilizando las herramientas del anotador. El anotador de Piximi ofrece varias opciones para **añadir**, **restar** o **interseccionar** anotaciones. Además, la **herramienta de selección** le permite **redimensionar** o **eliminar** anotaciones específicas. Para empezar a editar, seleccione imágenes específicas, o todas las imágenes, haciendo clic en la casilla de verificación de la parte superior.
-* Opcionalmente, puede ajustar los canales: Aunque hay dos canales en este experimento, la señal de los núcleos se duplicó en los canales rojo y verde. Este diseño está pensado para ser **color-blind friendly** y para producir un **color magenta** para los núcleos. El **canal verde** también incluye señales citoplasmáticas.
-
-Otra razón para duplicar los canales es que algunos modelos (como **Cellpose** que usamos hoy) requieren que las imágenes de entrada tengan **tres canales**.
-
-* Puede optar por segmentar manualmente las células para generar máscaras para los datos de 'verdad de referencia' (*ground truth*).
-
-##### 5. **Clasificar células**
-
-Razón para hacer esto: Queremos clasificar las “CELLPOSE_CELLS” basándonos en la distribución de la GFP (en Núcleos, citoplasma, o sin GFP) sin etiquetarlas todas y cada una manualmente. Para ello, podemos utilizar la función de clasificación en Piximi, que nos permite entrenar un clasificador utilizando un pequeño subconjunto de datos etiquetados y luego clasificar automáticamente las células restantes.
-
- 🔴 PARA HACER
-
-* Ir a la pestaña **CELLPOSE_CELLS** que muestra los objetos segmentados (flecha 1, figura 36).
-* Hacer clic en la pestaña **Clasificación** del panel izquierdo (flecha 2, figura 36).
-* Cree nuevas categorías haciendo clic en «+ Category». Añadir «Cytoplasmatic_GFP», «Nuclear_GFP», «No_GFP» tres categorías (flecha 3, figura 36).
-* Haga clic en las imágenes que coincidan con sus criterios. Puede seleccionar varias células manteniendo pulsado al tecla **Comamnd (⌘)** en Mac o **Shift** en Linux. Intenta asignar **~20-40 células por categoría**. Una vez seleccionadas, haz clic en **«Categorize»** para asignar las etiquetas a las células seleccionadas.
-
-```{figure} ./img/tutorial_images/Figure6.png
-:width: 600
-:align: center
-
-Clasificando células individuales en base a la presencia de GFP y su localización.
-```
-
-##### 6. **Entrenar el modelo clasificador**
-
- 🔴 PARA HACER
-
-* Haz clic en el icono « - Fit Model» para abrir la configuración de los hiperparámetros del modelo. Para el ejercicio de hoy, ajustaremos algunos parámetros:
-* Haga clic en «Architecture Settings» y ajuste la *Model Architecture* a **SimpleCNN**.
-* Actualice las dimensiones de entrada a
- - Filas de entrada: 48
- - Columnas de entrada: 48
- - Canales: 3 (ya que nuestras imágenes están en formato RGB)
-
- (Puede cambiar a otros números como 64, 128)
-
-```{figure} ./img/tutorial_images/Figure7.png
-:width: 600
-:align: center
-
-Configuración del modelo clasificador.
-```
-
-* Haga clic en la pestaña «Dataset Setting» y establezca el porcentaje de entrenamiento (*training percentage*) en 0,75, que reserva el 25% de los datos etiquetados para la validación.
-* Cuando haga clic en **"Fit Classifier"** en Piximi, aparecerán dos gráficos de entrenamiento "**Precisión vs Épocas "** y **"Pérdida vs Épocas "**. Cada gráfico muestra curvas para datos de **entrenamiento** y **validación**.
-* En el gráfico de **precisión**, verás lo bien que está aprendiendo el modelo. Lo ideal es que tanto la precisión de entrenamiento como la de validación aumenten y se mantengan cercanas.
-* En el gráfico de pérdidas, los valores más bajos significan un mejor rendimiento. Si la pérdida de validación empieza a aumentar mientras la pérdida de entrenamiento sigue cayendo, el modelo podría estar sobreajustándose.
-
-Estos gráficos le ayudan a comprender cómo está aprendiendo el modelo y si es necesario realizar ajustes.
-
-##### 7. **Evaluar el modelo:**
-
-🔴 PARA HACER
-
-```{figure} ./img/tutorial_images/Figure8.png
-:width: 400
-:align: center
-
-Entrenamiento y validación del clasificador.
-```
-
-* Haga clic en **«Predict model» (figura 38, flecha 1)** para aplicar el modelo que acabamos de entrenar. Este paso generará predicciones en las células que no hemos anotado.
-* Puede revisar las predicciones en la pestaña CELLPOSE_CELLS y eliminar cualquier categoría mal asignada.
-* Opcionalmente, puede seguir utilizando las etiquetas para refinar la verdad de referencia (*ground truth*) y mejorar el clasificador. Este proceso es parte de la clasificación **Human-in-the-loop**, donde se corrige iterativamente y entrenar el modelo basado en la entrada humana.
-* Haga clic en **«Evaluate model» (figura 38, flecha 2)** para evaluar el modelo que acabamos de entrenar. Las métricas de confusión y de evaluación pueden compararse con la verdad de referencia (*ground truth*).
-* Haga clic en «Accept Prediction (Hold)» (deberás mantener presionado el cursor unos segundos), para asignar las etiquetas predichas a todos los objetos.
-
-##### 8. **Medición**
-
-Una vez que esté satisfecho con la clasificación, procederemos a medir los objetos. El objetivo del ejercicio de hoy es determinar la concentración mínima de Wortmannin necesaria para bloquear la exportación de FOXO1A-GFP desde los núcleos. Para ello, podemos medir la intensidad total de GFP a nivel de imagen o a nivel de objeto.
-
-🔴 PARA HACER
-
-* Haga clic en «Measurement» en la esquina superior derecha.
-* Hacer clic en Tablas (Flecha 1) y seleccionar Imagen y hacer clic en «Confirm» (Flecha 2).
-* Elija «MEASUREMENT» en el panel izquierdo, tenga en cuenta que el paso de medición puede tardar algún tiempo en procesarse.
-* Haga clic en «Category» para incluir todas las categorías en la medición.
-* En «Total», haga clic en «Channell 1» (Flecha 3) para seleccionar la medición para GFP. Verá la medición en la pestaña «DATA GRID». Las mediciones se presentan como valores medios o medianos, y el conjunto de datos completo está disponible al exportar el archivo .csv.
-
-```{figure} ./img/tutorial_images/Figure9.png
-:width: 600
-:align: center
-
-Agregar mediciones.
-```
-
-##### 9. **Visualización**
-
-Después de generar las mediciones, puede trazar las mediciones.
-
-🔴 PARA HACER
-
-* Haga clic en “PLOTS” (Figura 40, Flecha 1) para visualizar las mediciones.
-* Establezca el tipo de trazado en “Swarm” y elija un tema de color basado en su preferencia.
-* Seleccione “Y-axis” como “intensity-total-channel-1” y establezca “SwarmGroup” como “category”; esto generará una curva mostrando cómo varía la intensidad de GFP a través de diferentes categorías (Figura 40, Flecha 2).
-* Seleccionando “Show Statistics” se mostrará la media, así como los límites de confianza superior e inferior, en el gráfico.
-* Opcionalmente, puede experimentar con diferentes tipos de gráficos y ejes para ver si los datos revelan información adicional.
-
-```{figure} ./img/tutorial_images/Figure10.png
-:width: 600
-:align: center
-
-Graficar los resultados.
-```
-
-##### 10. **Exportar los resultados y guardar el proyecto**
-
-🔴 PARA HACER
-
-* Haz clic en «SAVE» en la esquina superior izquierda para guardar todo el proyecto. Verás la animación del logo de Piximi a medida que avanza el guardado .
-
-##### 11. **Información adicional**
-
-Consulta el paper de Piximi: [https://www.biorxiv.org/content/10.1101/2024.06.03.597232v2](https://www.biorxiv.org/content/10.1101/2024.06.03.597232v2)
-
-Consulta la documentación de Piximi:[Documentación de Piximi](https://documentation.piximi.app/intro.html):[https://documentation.piximi.app/intro.html](https://documentation.piximi.app/intro.html)
-
-Informar de fallos/errores o solicitar características [https://github.com/piximi/documentation/issues](https://github.com/piximi/documentation/issues)
\ No newline at end of file
diff --git a/piximi-documentation/translocation_tutorial_pt_BR.md b/piximi-documentation/translocation_tutorial_pt_BR.md
deleted file mode 100644
index 0a08937..0000000
--- a/piximi-documentation/translocation_tutorial_pt_BR.md
+++ /dev/null
@@ -1,230 +0,0 @@
-# Tutorial para iniciantes do Piximi (Portugues)
-
-## Segmentação e classificação sem instalação no navegador
-
-Beth Cimini, Le Liu, Esteban Miglietta, Paula Llanos, Nodar Gogoberidze
-
-Instituto Broad do MIT e Harvard, Cambridge, MA.
-
-### **Informações básicas:**
-
-#### **O que é Piximi?**
-
-Piximi é uma ferramenta moderna de análise de imagens sem programação que utiliza aprendizado profundo. Implementado como um aplicativo web em [https://piximi.app/](https://piximi.app/), o Piximi não requer instalação e pode ser acessado por qualquer navegador moderno. Sua arquitetura exclusiva para clientes preserva a segurança dos dados do pesquisador, executando toda a computação localmente.
-
-O Piximi é interoperável com ferramentas e fluxos de trabalho existentes, suportando importação e exportação de dados e formatos de modelos comuns. A interface intuitiva e o fácil acesso ao Piximi permitem que pesquisadores obtenham insights sobre imagens em apenas alguns minutos. O Piximi visa levar a análise de imagens com aprendizado profundo a uma comunidade mais ampla, eliminando barreiras.
-
-
-\* exceto as segmentações usando Cellpose, que são enviadas para um servidor remoto (com a permissão do usuário).
-
-Funcionalidades principais: **Anotador, Segmentador, Classificador, Medições.**
-
-#### **Objetivo do exercício**
-
-Neste exercício, você se familiarizará com as principais funcionalidades do Piximi: anotação, segmentação, classificação, mensuração e visualização, e o utilizará para analisar um conjunto de imagens de um experimento de translocação. O objetivo deste experimento é determinar a **menor dose efetiva** de Wortmannin necessária para induzir a localização nuclear de FOXO1A marcada com GFP (Figura 21). Você segmentará as imagens usando um dos modelos de aprendizado profundo disponíveis no Piximi, verificará e selecionará a segmentação e, em seguida, treinará um classificador de imagens para classificar as células individuais como tendo "GFP nuclear", "GFP citoplasmática" ou "sem GFP". Por fim, você fará medições e as plotará para responder à pergunta biológica.
-
-
-#### **Contexto do experimento**
-
-Neste experimento, pesquisadores obtiveram imagens de células U2OS de osteossarcoma (câncer ósseo) fixadas expressando uma proteína de fusão FOXO1A-GFP e coraram DAPI para marcar os núcleos. FOXO1 é um fator de transcrição que desempenha um papel fundamental na regulação da gliconeogênese e glicogenólise por meio da sinalização da insulina. FOXO1A transita dinamicamente entre o citoplasma e o núcleo em resposta a vários estímulos. A wortmanina, um inibidor da PI3K, pode bloquear a exportação nuclear, resultando no acúmulo de FOXO1A no núcleo.
-
-
-
-```{figure} ./img/tutorial_images/Figure1.png
-:largura: 300
-:alinhar: centro
-
-Representação esquemática do mecanismo FOXO1A
-```
-
-#### **Materiais necessários para este exercício**
-
-Os materiais necessários para este exercício podem ser baixados de: [PiximiTutorial](./downloads/Piximi_Translocation_Tutorial_RGB.zip). O arquivo “Piximi Translocation Tutorial RGB.zip” contém um projeto Piximi, incluindo todas as imagens, já rotuladas com o tratamento correspondente (concentração de Wortmannin ou Controle). Baixe este arquivo, mas **NÃO o descompacte**!
-
-#### **Instruções do exercício**
-
-Leia os passos abaixo e siga as instruções onde indicado. Os passos em que você precisa encontrar uma solução estão marcados com 🔴 PARA FAZER.
-
-##### 1. **Carregue o projeto Piximi**
-
-🔴 PARA FAZER
-
-* Inicie o Piximi acessando: [https://piximi.app/](https://piximi.app/)
-
-* Carregue o projeto de exemplo: Clique em “Abrir” \- “Projeto” \- “Projeto do Zip”, como mostrado na figura 22, para carregar um arquivo de projeto para este tutorial do Zip. Você também pode alterar o nome do projeto no painel superior esquerdo, como “Exercício Piximi”. Conforme ele é carregado, você pode ver a progressão no logotipo no canto superior esquerdo.
-
-```{figure} ./img/tutorial_images/Figure2.png
-:width: 600
-:align: center
-
-Carregando um arquivo de projeto.
-```
-
-##### 2. **Verifique as imagens carregadas e explore a interface do Piximi**
-
-Estas 17 imagens representam tratamentos com Wortmannin em oito concentrações diferentes (expressas em nM), bem como tratamentos controles (0 nM). Observe que o canal DAPI (Núcleos) é mostrado em magenta e que o canal GFP (FOXOA1) é mostrado em verde.
-
-Ao passar o mouse sobre a imagem, rótulos coloridos são exibidos no canto esquerdo das imagens. Essas anotações são dos metadados do arquivo compactado que acabamos de enviar. Neste tutorial, os diferentes rótulos coloridos indicam a concentração de Wortmannin, enquanto os números representam o número de imagens em cada categoria.
-
-Opcionalmente, você pode anotar as imagens manualmente clicando em "+ Categoria", inserindo seu rótulo e, em seguida, selecionando a imagem clicando nas imagens e anotando as imagens selecionadas clicando em **"Categorizar"**. Neste tutorial, pularemos esta etapa, pois os rótulos já foram carregados no início.
-
-```{figure} ./img/tutorial_images/Figure3.png
-:largura: 600
-:alinhar: centro
-
-Explorando as imagens e rótulos.
-```
-
-##### 3. **Segmentar Células - descubra as células a partir do fundo**
-
-🔴 PARA FAZER
-
-* Para iniciar a previsão em todas as imagens, clique em “Selecionar Todas as Imagens” no painel superior, conforme mostrado na Figura 23.
-* Altere a Tarefa de Aprendizagem para “SEGMENTAÇÃO” (Figura 24, Seta 1).
-
-* Clique em “+ CARREGAR MODELO” (Seta 2) e a janela será exibida, permitindo que você escolha um modelo pré-treinado (Seta 3). Para o exercício de hoje, selecione “Cellpose” (Seta 4). Mais informações sobre o modelo suportado podem ser encontradas [aqui](https://documentation.piximi.app/segmentation.html).
-* Clique em “Abrir Modelo de Segmentação” (Seta 5) para carregar seu modelo e selecioná-lo. Por fim, clique em “Prever Modelo” (Seta 5). Você verá o progresso da previsão exibido no canto superior esquerdo, abaixo do logotipo da Piximi .
-* A segmentação levará alguns minutos para ser concluída.
-
-```{figure} ./img/tutorial_images/Figure4.png
-:width: 600
-:align: center
-
-Carregando um modelo de segmentação.
-```
-
-Observe que as etapas anteriores foram executadas em sua máquina local, o que significa que suas imagens estão armazenadas localmente. No entanto, a inferência do Cellpose é executada na nuvem, o que significa que suas imagens serão enviadas para processamento. Se suas imagens forem altamente sensíveis, tenha cuidado ao usar serviços baseados em nuvem.
-
-##### 4. **Visualize o resultado da segmentação e corrija os erros de segmentação**
-
-🔴 PARA FAZER
-
-* Clique na aba **CELLPOSE_CELLS** para verificar as células individuais que foram segmentadas. Clique na aba “IMAGEM” e depois em “Anotar” para verificar a segmentação de toda a imagem.
-
-```{figure} ./img/tutorial_images/Figure5.png
-:largura: 600
-:alinhar: centro
-
-Ferramenta de anotação do Piximi.
-```
-
-* Opcionalmente, aqui você pode refinar manualmente a segmentação usando as ferramentas do anotador. O anotador Piximi oferece diversas opções para **adicionar**, **subtrair** ou **interseccionar** anotações. Além disso, a **ferramenta de seleção** permite **redimensionar** ou **excluir** anotações específicas. Para começar a editar, selecione imagens específicas ou todas clicando na caixa de seleção na parte superior.
-* Opcionalmente, você pode ajustar os canais: embora existam dois canais neste experimento, o sinal dos núcleos é duplicado nos canais vermelho e verde. Este projeto foi projetado para ser **compatível com daltonismo** e produzir uma **cor magenta** para os núcleos. O **canal verde** também inclui sinais citoplasmáticos.
-
-Outro motivo para duplicar os canais é que alguns modelos — como o **modelo Cellpose** que usamos hoje — exigem uma entrada de **três canais**.
-
-* Você pode optar por segmentar manualmente as células para gerar máscaras para dados de verdade básica.
-
-##### **Classificar células**
-
-Motivo para isso: Queremos classificar as 'CELLPOSE\_CELLS' com base na distribuição de GFP (em núcleos, citoplasma ou sem GFP) sem rotular todas elas manualmente. Para isso, podemos usar a função de classificação do Piximi, que nos permite treinar um classificador usando um pequeno subconjunto de dados rotulados e, em seguida, classificar automaticamente as células restantes.
-
-🔴 PARA FAZER
-
-* Acesse a aba **CELLPOSE_CELLS** que exibe os objetos segmentados (seta 1, figura 26)
-* Clique na aba **Classificação** no painel esquerdo (seta 2, figura 26).
-* Crie novas categorias clicando em **“+ Categoria”**. Adicione as três categorias “Cytoplasmatic_GFP”, “Nuclear_GFP” e “No_GFP” (Seta 3, Figura 26).
-* Clique nas imagens que correspondem aos seus critérios. Você pode selecionar várias células pressionando **Command (⌘)** no Mac ou **Shift** no Linux. Tente atribuir **\~20–40 células por categoria**. Após selecionar, clique em **“Categorizar”** para atribuir os rótulos às células selecionadas.
-
-```{figure} ./img/tutorial_images/Figure6.png
-:width: 600
-:align: center
-
-Classificando células individuais com base na presença e localização de GFP.
-```
-
-##### 6. **Treine o modelo do Classificador**
-
-🔴 PARA FAZER
-
-* Clique no ícone " - Ajustar Modelo" para abrir as configurações de hiperparâmetros do modelo. Para o exercício de hoje, ajustaremos alguns parâmetros:
-* Clique em “Configurações de arquitetura” e defina a arquitetura do modelo como **SimpleCNN**.
-* Atualize as dimensões de entrada para:
- - Linhas de entrada: 48
- - Colunas de entrada: 48
- - Canais: 3 (já que nossas imagens estão no formato RGB)
-
- (Você pode mudar para outros números, como 64, 128)
-
-```{figure} ./img/tutorial_images/Figure7.png
-:largura: 600
-:alinhar: centro
-
-Configuração do modelo classificador.
-```
-
-* Clique na aba “Configuração do Conjunto de Dados” e defina a Porcentagem de Treinamento como 0,75, o que reserva 25% dos dados rotulados para validação.
-* Ao clicar em **"Ajustar Classificador"** no Piximi, dois gráficos de treinamento aparecerão: "**Precisão vs. Épocas'' e **"Perda vs. Épocas''. Cada gráfico mostra curvas para os dados de **treinamento** e **validação**.
-* No **gráfico de precisão**, você verá o quão bem o modelo está aprendendo. Idealmente, a precisão tanto do treinamento quanto da validação deve aumentar e permanecer próxima.
-* No **gráfico de perdas**, valores menores significam melhor desempenho. Se a perda de validação começar a aumentar enquanto a perda de treinamento continua caindo, o modelo pode estar com sobreajuste.
-
-Esses gráficos ajudam a entender como o modelo está aprendendo e se ajustes são necessários.
-
-##### 7. **Avaliar modelo:**
-
-🔴 A FAZER
-
-```{figure} ./img/tutorial_images/Figure8.png
-:width: 400
-:align: center
-
-Treinamento e validação do classificador.
-```
-
-* Clique em **“Prever Modelo” (figura 28, seta 1)** para aplicar o modelo que acabamos de treinar. Esta etapa gerará previsões nas células que não anotamos.
-* Você pode revisar as previsões na guia CELLPOSE_CELLS e excluir quaisquer categorias atribuídas incorretamente.
-* Opcionalmente, você pode continuar usando os rótulos para refinar a verdade básica e aprimorar o classificador. Esse processo faz parte da **classificação humana no ciclo**, na qual você corrige e treina o modelo iterativamente com base na entrada humana.
-* Clique em **“Avaliar Modelo” (figura 28, seta 2)** para avaliar o modelo que acabamos de treinar. As métricas de confusão e de avaliação podem ser comparadas com a verdade básica.
-* Clique em "Aceitar previsão (Manter)" para atribuir os rótulos previstos a todos os objetos.
-
-##### 8. **Medição**
-
-Assim que estiver satisfeito com a classificação, prosseguiremos com a medição dos objetos. O objetivo do exercício de hoje é determinar a concentração mínima de Wortmannin necessária para bloquear a exportação de FOXO1A-GFP dos núcleos. Para isso, podemos medir a intensidade total de GFP na imagem ou no objeto.
-
-🔴 PARA FAZER
-
-* Clique em “Medição” no canto superior direito.
-* Clique em Tabelas (Seta 1), selecione Imagem e clique em “Confirmar” (Seta 2).
-* Selecione "MEDIÇÃO" no painel esquerdo. Observe que a etapa de medição pode levar algum tempo para ser processada.
-* Clique em "Categoria" para incluir todas as categorias na medição.
-* Em "Total", clique em "Canal 1" (Seta 3) para selecionar a medição para GFP. Você verá a medição na aba "GRADE DE DADOS". As medições são apresentadas como valores médios ou medianos, e o conjunto de dados completo está disponível ao exportar o arquivo .csv.
-
-```{figure} ./img/tutorial_images/Figure9.png
-:largura: 600
-:alinhar: centro
-
-Adicione medidas.
-```
-
-##### 9. **Visualização**
-
-Após gerar as medições, você pode plotá-las.
-
-🔴 PARA FAZER
-
-* Clique em "PLOTS" (Figura 30, Seta 1) para visualizar as medições.
-* Defina o tipo de gráfico como "Swarm" e escolha um tema de cores de acordo com sua preferência.
-* Selecione "Y-axis" como "intensity-total-channel-1" e defina "SwarmGroup" como "category"; isso gerará uma curva mostrando como a intensidade da GFP varia entre as diferentes categorias (Figura 30, Seta 2).
-* Selecionar "Show Statistics" exibirá a média, bem como os limites de confiança superior e inferior, no gráfico.
-* Opcionalmente, você pode experimentar diferentes tipos de gráfico e eixos para ver se os dados revelam insights adicionais.
-
-```{figure} ./img/tutorial_images/Figure10.png
-:width: 600
-:align: center
-
-Resultados do gráfico.
-```
-
-##### 10. **Exportar resultados e salvar o projeto**
-
-🔴 PARA FAZER
-
-* Clique em "SALVAR" no canto superior esquerdo para salvar o projeto inteiro. Você verá a animação do logotipo do Piximi conforme o salvamento avança .
-
-##### 11. **Informações de apoio**
-
-Confira o artigo do Piximi: [https://www.biorxiv.org/content/10.1101/2024.06.03.597232v2](https://www.biorxiv.org/content/10.1101/2024.06.03.597232v2)
-
-Confira a documentação do Piximi:[Documentação do Piximi](https://documentation.piximi.app/intro.html):[https://documentation.piximi.app/intro.html](https://documentation.piximi.app/intro.html)
-
-Relatar bugs/erros ou solicitar recursos [https://github.com/piximi/documentation/issues](https://github.com/piximi/documentation/issues)
Launch Piximi here
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+