update documentation

CHANGELOG.md

@@ -6,7 +6,7 @@
 ## Release 2.0.0
 * train models with nnU-Net v2 (nnunet_cust dependency no longer needed)
 * roi_subset a lot faster, because cropping with 6mm low res model to roi first
-* improved training dataset (for details see `resources/improvements_in_v2.md`)
+* more classes and improved training dataset (for details see `resources/improvements_in_v2.md`)
 
 
 ## Release 1.5.6

README.md

@@ -2,7 +2,7 @@
 
 Tool for segmentation of 104 classes in CT images. It was trained on a wide range of different CT images (different scanners, institutions, protocols,...) and therefore should work well on most images. The training dataset with 1204 subjects can be downloaded from [Zenodo](https://doi.org/10.5281/zenodo.6802613). You can also try the tool online at [totalsegmentator.com](https://totalsegmentator.com/).
 
-Overview of [improvements](resources/improvements_in_v2.md) in TotalSegmentator v2.
+Overview of [changes and improvements](resources/improvements_in_v2.md) in TotalSegmentator v2.
 
 ![Alt text](resources/imgs/overview_classes.png)
 

resources/improvements_in_v2.md

@@ -1,22 +1,49 @@
-# Improvements in training data of TotalSegmentator v2
+# Changes and improvements in TotalSegmentator v2
 
-List of new classes:
-* costal cartilages
-* prostate
-* thyroid
-* TODO: add all
+## New classes (n=33)
 
-Some of these new classes were available in some preliminary version as additional models in v1. Now they are properly added.
+List of new classes by task:
 
+total:
+```
+skull, thyroid_gland, prostate, brachiocephalic_vein_left, brachiocephalic_vein_right, brachiocephalic_trunk, common_carotid_artery_left, common_carotid_artery_right, atrial_appendage_left, subclavian_artery_left, subclavian_artery_right, vertebrae_S1, sternum, costal_cartilages, pulmonary_vein, superior_vena_cava, kidney_cyst_left, kidney_cyst_right, spinal_cord
+```
+
+appendicular_bones: 
+```
+patella, tibia, fibula, tarsal, metatarsal, phalanges_feet, ulna, radius, carpal, metacarpal, phalanges_hand
+```
+
+tissue_types:
+```
+subcutaneous_fat, skeletal_muscle, torso_fat
+```
+
+vertebrae_body:
+```
+vertebrae_body (vertebrae without the spinous process)
+```
+
+The following classes were moved from the `total` task to the `heartchambers_highres` task:
+```
+heart_myocardium, heart_atrium_left, heart_ventricle_left, heart_atrium_right, heart_ventricle_right, pulmonary_artery
+```
+`total` now only contains the overal class `heart` instead.
+
+Some of these new classes were available in some preliminary version as additional tasks in v1. Now they are properly added.
+
+
+## Improvements in training dataset
 
 List of classes where we corrected some systemic errors in the labels (e.g. myocardium always slightly misaligned) or corrected several errors (e.g. bleedings in the liver sometimes not correctly labeled). For these classes you can expect a bit better segmentations now:
 * femur
+* humerus
 * hip
 * heart chambers
 * aorta
 * liver
 * spleen
-
+* kidney
 
 We increased the number of training images from 1139 to 1559. We added the following images:
 * more whole body images where TotalSegmentator failed before 
@@ -25,6 +52,8 @@ We increased the number of training images from 1139 to 1559. We added the follo
 * more images with bleedings in the abdomen where TotalSegmentator failed before
 * more images from GE scanners and other institutions
 
+> NOTE: The public dataset does not contain these additional subjects. However, it contains several improvements (see next section).
+
 
 Still open problems:
 * the start of the ribs close to the spine is always missing a small part
@@ -34,7 +63,8 @@ Still open problems:
 
 
 ## Changes/improvements for public dataset:
-* XX new classes (all classes of the `total` task)
+* increased number of classes from 104 to 117 (all classes of the `total` task)
+* removed heart chamber classes
 * improved label quality (see above) (better than before but probably still some errors)
 * more meta data per image (scanner, pathology, ...)
 * less intrusive defacing

totalsegmentator/map_to_binary.py

@@ -254,6 +254,9 @@
         1: "liver_vessels",
         2: "liver_tumor"
     },
+    "vertebrae_body": {
+        1: "vertebrae_body"
+    },
     "heartchambers_highres": {
         1: "heart_myocardium", 
         2: "heart_atrium_left",