Spaces:
Running
Running
| # Getting Started | |
| ## Installation | |
| For installation and downloading the saved weights of the models, please check the [github](https://github.com/anirudhbalaraman/WSAttention-Prostate) repository. | |
| ## Data Format | |
| Input bpMRI scans should have all three sequnces, T2W, ADC and DWI as separate files in **NRRD** format. The prostate mask and heatmap are generated by the sample for each sample. To train the models or run testing with reference labels, the data pipeline uses MONAI's decathlon-format JSON: | |
| ```json | |
| { | |
| "train": [ | |
| { | |
| "image": "path/to/t2.nrrd", | |
| "dwi": "path/to/dwi.nrrd", | |
| "adc": "path/to/adc.nrrd", | |
| "mask": "path/to/prostate_mask.nrrd", | |
| "heatmap": "path/to/heatmap.nrrd", | |
| "label": 0 | |
| } | |
| ], | |
| "test": [...] | |
| } | |
| ``` | |
| Sample JSON files are available in the github repository. | |
| Paths are relative to `data_root`. PI-RADS labels are 0-indexed (`0` = PI-RADS 2, `3` = PI-RADS 5). csPCa labels are binary (0 or 1). | |