Make compatible with latest DiffDRR (#20)

* Fix fiducials in DeepFluoro

* Bump version

* Simplify transforms

* Separate import and code cells

* Update README.md

* Pin DiffDRR version

Author: eigenvivek

README.md

@@ -79,7 +79,7 @@ from diffdrr.visualization import plot_drr
 from diffdrrdata.deepfluoro import DeepFluoroDataset, Transforms
 
 # Load a subject from the DeepFluoroDataset
-deepfluoro = DeepFluoroDataset(id_number=1, bone_attenuation_multiplier=2.5)
+deepfluoro = DeepFluoroDataset(id_number=1)
 
 # Initialize the DRR module
 subsample = 4
@@ -142,7 +142,6 @@ drr = DRR(
     dely * subsample,
     x0=x0,
     y0=y0,
-    reverse_x_axis=False,
 )
 transform = Transforms(height // subsample, width // subsample)
 

diffdrrdata/__init__.py

@@ -1 +1 @@
-__version__ = "0.0.4"
+__version__ = "0.0.5"

diffdrrdata/deepfluoro.py

@@ -9,15 +9,14 @@
 import h5py
 import numpy as np
 import torch
+from .utils import load_file
 from torchio import LabelMap, ScalarImage, Subject
 from torchio.transforms.preprocessing import ToCanonical
 from torchvision.transforms.functional import center_crop, gaussian_blur
 
 from diffdrr.data import read
+from diffdrr.detector import parse_intrinsic_matrix
 from diffdrr.pose import RigidTransform
-from diffdrr.utils import parse_intrinsic_matrix
-
-from .utils import load_file
 
 # %% ../notebooks/00_deepfluoro.ipynb 6
 class DeepFluoroDataset(torch.utils.data.Dataset):
@@ -64,8 +63,8 @@ def __init__(
         self.flip_z = RigidTransform(
             torch.tensor(
                 [
-                    [0, 1, 0, 0],
                     [-1, 0, 0, 0],
+                    [0, -1, 0, 0],
                     [0, 0, -1, 0],
                     [0, 0, 0, 1],
                 ]
@@ -94,12 +93,10 @@ def __getitem__(self, idx):
         pose = self.projections[f"{idx:03d}/gt-poses/cam-to-pelvis-vol"][:]
         pose = RigidTransform(torch.from_numpy(pose))
         pose = (
-            self.rot_180
-            .compose(self.flip_z)
+            self.flip_z
             .compose(self.world2camera.inverse())
             .compose(pose)
             .compose(self.anatomical2world)
-            .compose(self.rot_180)
         )
         if self.rot_180_for_up(idx):
             img = torch.rot90(img, k=2)
@@ -121,11 +118,11 @@ def parse_volume(subject, bone_attenuation_multiplier, labels):
     # Get all parts of the volume
     volume = subject["vol/pixels"][:]
     volume = np.swapaxes(volume, 0, 2).copy()
-    volume = torch.from_numpy(volume).unsqueeze(0).flip(1).flip(2)
+    volume = torch.from_numpy(volume).unsqueeze(0)  # .flip(1).flip(2)
 
     mask = subject["vol-seg/image/pixels"][:]
     mask = np.swapaxes(mask, 0, 2).copy()
-    mask = torch.from_numpy(mask).unsqueeze(0).flip(1).flip(2)
+    mask = torch.from_numpy(mask).unsqueeze(0)  # .flip(1).flip(2)
 
     affine = np.eye(4)
     affine[:3, :3] = subject["vol/dir-mat"][:]
@@ -150,10 +147,10 @@ def parse_volume(subject, bone_attenuation_multiplier, labels):
     anatomical2world = RigidTransform(
         torch.tensor(
             [
-                [1.0, 0.0, 0.0, -isocenter[0]],
-                [0.0, 1.0, 0.0, -isocenter[1]],
-                [0.0, 0.0, 1.0, -isocenter[2]],
-                [0.0, 0.0, 0.0, 1.0],
+                [1, 0, 0, -isocenter[0]],
+                [0, 1, 0, -isocenter[1]],
+                [0, 0, 1, -isocenter[2]],
+                [0, 0, 0, 1],
             ],
             dtype=torch.float32,
         )
@@ -169,24 +166,22 @@ def parse_volume(subject, bone_attenuation_multiplier, labels):
         label_def=defns,
         fiducials=fiducials,
     )
-    reorient = RigidTransform(torch.diag(torch.tensor([-1.0, -1.0, 1.0, 1.0])))
-    subject.fiducials = reorient(subject.fiducials)
 
     return subject, anatomical2world
 
 
 def parse_proj_params(f):
     proj_params = f["proj-params"]
     extrinsic = torch.from_numpy(proj_params["extrinsic"][:])
-    world2camera = RigidTransform(extrinsic)
+    camera2world = RigidTransform(extrinsic)
     intrinsic = torch.from_numpy(proj_params["intrinsic"][:])
     num_cols = proj_params["num-cols"][()]
     num_rows = proj_params["num-rows"][()]
     proj_col_spacing = float(proj_params["pixel-col-spacing"][()])
     proj_row_spacing = float(proj_params["pixel-row-spacing"][()])
     return (
         intrinsic,
-        world2camera,
+        camera2world,
         num_cols,
         num_rows,
         proj_col_spacing,
@@ -200,7 +195,7 @@ def load(id_number, bone_attenuation_multiplier, labels):
     # Load dataset parameters
     (
         intrinsic,
-        world2camera,
+        camera2world,
         num_cols,
         num_rows,
         proj_col_spacing,
@@ -227,13 +222,15 @@ def load(id_number, bone_attenuation_multiplier, labels):
     ][id_number - 1]
     subject = f[subject_id]
     projections = subject["projections"]
-    subject, anatomical2world = parse_volume(subject, bone_attenuation_multiplier, labels)
+    subject, anatomical2world = parse_volume(
+        subject, bone_attenuation_multiplier, labels
+    )
 
     return (
         subject,
         projections,
         anatomical2world,
-        world2camera,
+        camera2world,
         focal_len,
         int(num_rows),
         int(num_cols),

diffdrrdata/ljubljana.py

@@ -35,8 +35,8 @@ def __init__(
         self.flip_z = RigidTransform(
             torch.tensor(
                 [
-                    [0, 1, 0, 0],
-                    [1, 0, 0, 0],
+                    [-1, 0, 0, 0],
+                    [0, -1, 0, 0],
                     [0, 0, -1, 0],
                     [0, 0, 0, 1],
                 ]
@@ -118,7 +118,7 @@ def parse_volume(f, subject_id):
     return subject, anatomical2world
 
 # %% ../notebooks/01_ljubljana.ipynb 7
-from diffdrr.utils import parse_intrinsic_matrix
+from diffdrr.detector import parse_intrinsic_matrix
 
 
 def parse_proj(f, subject_id, view):