Add comprehensive AI methods documentation

Create VisCy-style documentation for CAREamics and Spotiflow, and reorganize
README with prominent AI methods section.

New Documentation:
- docs/careamics_denoising.md: Complete guide to CAREamics denoising
  * Noise2Void, CARE, N2V2, Struct N2V overview
  * Training and usage instructions
  * Parameter explanations and tuning guidelines
  * Hardware requirements and performance tips
  * Troubleshooting section
  * Example workflows and pipelines

- docs/spotiflow_detection.md: Complete guide to Spotiflow spot detection
  * Pretrained models overview (general, hybiss, synth_3d, etc.)
  * 2D and 3D spot detection
  * Parameter tuning for different scenarios
  * smFISH analysis examples
  * Performance optimization
  * Custom model training guidance

README Updates:
- Add prominent '🤖 AI-Powered Processing Functions' section
- Reorganize documentation links by category
- Highlight key capabilities of each AI method
- Better visual structure with emojis and clear grouping
- Link all 5 AI methods (VisCy, CAREamics, Spotiflow, Cellpose, Trackastra)

Documentation follows VisCy template:
✓ Overview and requirements
✓ Model information and references
✓ Installation and setup
✓ Basic and advanced usage examples
✓ Complete parameter documentation
✓ Typical workflows
✓ Hardware recommendations
✓ Comprehensive troubleshooting
✓ Tips and best practices
✓ Citations and resources

Author: macromeer

README.md

@@ -94,32 +94,50 @@ Converts `.lif, .nd2, .czi, .ndpi` and Acquifer data to TIF or OME-Zarr formats.
 Note that whenever you click on an `Original File` or `Processed File` in the table, it will replace the one that is currently shown in the viewer. So naturally, you'd first select the original image, and then the processed image to correctly see the image pair that you want to inspect.
 
 
-#### Processing Function Credits
-
-The image processing capabilities are powered by several excellent open-source tools:
-- [Cellpose 4](https://github.com/MouseLand/cellpose): Advanced cell segmentation
-- [Trackastra](https://github.com/weigertlab/trackastra): Cell tracking and analysis
-- [VisCy](https://github.com/mehta-lab/VisCy): Virtual staining using deep learning
-- [CAREamics](https://github.com/CAREamics/careamics): Content-aware image restoration and enhancement
-- [Spotiflow](https://github.com/weigertlab/spotiflow): Accurate and efficient spot detection for fluorescence microscopy
-
-#### Processing Function Documentation
-
-Detailed documentation for specific processing functions:
-
-**Core Processing**
+#### AI-Powered Processing Functions 🤖
+
+napari-tmidas integrates state-of-the-art deep learning methods for microscopy image analysis. Each method runs in a dedicated environment to prevent conflicts and is automatically installed on first use.
+
+**🔬 Image Enhancement & Restoration**
+- **[VisCy Virtual Staining](docs/viscy_virtual_staining.md)** - Transform phase/DIC microscopy into virtual fluorescence (nuclei + membrane channels)
+  - Powered by [VisCy](https://github.com/mehta-lab/VisCy)
+  - VSCyto3D model for 3D/4D data
+  - No labels needed - learn from transmitted light alone
+  
+- **[CAREamics Denoising](docs/careamics_denoising.md)** - Content-aware image restoration using Noise2Void/CARE
+  - Powered by [CAREamics](https://github.com/CAREamics/careamics)
+  - Train on noisy images (no ground truth needed)
+  - Preserves fine structures while removing noise
+
+**🎯 Detection & Segmentation**
+- **[Spotiflow Spot Detection](docs/spotiflow_detection.md)** - Accurate detection of spots/puncta in fluorescence images
+  - Powered by [Spotiflow](https://github.com/weigertlab/spotiflow)
+  - Multiple pretrained models (2D/3D)
+  - Subpixel localization for precise coordinates
+  
+- **[Cellpose Segmentation](docs/cellpose_segmentation.md)** - Universal cell/nucleus segmentation
+  - Powered by [Cellpose 4](https://github.com/MouseLand/cellpose)
+  - Cellpose-SAM for improved generalization
+  - Works on 2D, 3D, and time-lapse data
+
+**🔍 Tracking & Analysis**
+- **[Trackastra Tracking](docs/trackastra_tracking.md)** - Cell tracking across time-lapse data
+  - Powered by [Trackastra](https://github.com/weigertlab/trackastra)
+  - Graph-based tracking with deep learning
+  - Handles divisions and complex cell behavior
+
+#### Additional Processing Functions
+
+**Core Operations**
 - [Basic Processing Functions](docs/basic_processing.md) - Label and intensity operations, channel splitting/merging, time series
-- [Cellpose Segmentation](docs/cellpose_segmentation.md) - Deep learning cell/nucleus segmentation
-- [TrackAstra Tracking](docs/trackastra_tracking.md) - Cell tracking across time-lapse data
-- [VisCy Virtual Staining](docs/viscy_virtual_staining.md) - Virtual staining of phase/DIC images using deep learning
 
-**Analysis and Quality Control**
+**Analysis & Quality Control**
 - [Grid View: Intensity + Labels Overlay](docs/grid_view_overlay.md) - Visual QC for segmentation results
 - [Intensity-Based Label Filtering](docs/intensity_label_filter.md) - Filter labels by signal intensity
 - [Regionprops Analysis](docs/regionprops_analysis.md) - Extract quantitative properties from labels
 
-**Advanced Processing**
-- [Advanced Processing Functions](docs/advanced_processing.md) - Denoising (CAREamics), spot detection (Spotiflow), SciPy/scikit-image filters, compression, colocalization
+**Advanced Methods**
+- [Advanced Processing Functions](docs/advanced_processing.md) - SciPy/scikit-image filters, compression, colocalization
 
 ### Batch Label Inspection
 If you have already segmented a folder full of images and now you want to maybe inspect and edit each label image, you can use the `Plugins > T-MIDAS > Batch Label Inspection`, which automatically saves your changes to the existing label image once you click the `Save Changes and Continue` button (bottom right).