Almost final verification code

spleenseg/core/neuralnet.py

@@ -42,8 +42,8 @@
 # from monai.data.dataset import Dataset
 from monai.data.utils import decollate_batch
 from monai.data.meta_tensor import MetaTensor
+from monai.handlers.utils import from_engine
 
-# from monai.config.deviceconfig import print_config
 # from monai.apps.utils import download_and_extract
 import torch
 
@@ -52,7 +52,7 @@
 # import shutil
 # import glob
 # import pudb
-from typing import Any
+from typing import Any, Sequence
 import numpy as np
 
 from spleenseg.transforms import transforms
@@ -296,7 +296,7 @@ def train(
         )
 
     def inference_metricsProcess(self) -> float:
-        metric: float = self.network.dice_metric.aggregate().item()
+        metric: float = self.network.dice_metric.aggregate().item()  # type: ignore
         self.trainingLog.metric_per_epoch.append(metric)
         self.network.dice_metric.reset()
         if metric > self.trainingLog.best_metric:
@@ -408,23 +408,38 @@ def plot_bestModel(
         )
         return 0.0
 
+    def bestModel_runOverValidationSpace(self):
+        self.network.model.load_state_dict(
+            torch.load(str(self.trainingParams.modelPth))
+        )
+        self.slidingWindowInference_do(self.validationSpace, self.plot_bestModel)
+
     def diceMetric_onValidationSpacing(
         self,
         sample: dict[str, MetaTensor | torch.Tensor],
         space: data.LoaderCache,
         index: int,
         result: torch.Tensor,
     ) -> float:
+        metric: float = -1.0
         sample["pred"] = result
         sample = [
             self.f_outputPost(i)
             for i in decollate_batch(sample)  # type: ignore[arg-type]
         ]
-        self.network.model.load_state_dict(
-            torch.load(str(self.trainingParams.modelPth))
+        predictions: torch.Tensor
+        labels: torch.Tensor
+        predictions, labels = from_engine(["pred", "label"])(sample)
+        Dm: torch.Tensor = self.network.dice_metric(
+            y_pred=predictions,  # type: ignore
+            y=labels,  # type: ignore
         )
-        self.slidingWindowInference_do(self.validationSpace, self.plot_bestModel)
-        return 0.0
+        print(f"Best prediction dice metric: {Dm}")
+        if space.loader.batch_size:
+            if index == len(space.cache) // space.loader.batch_size:
+                metric = self.network.dice_metric.aggregate().item()
+                print(f"metric on original image spacing: {metric}")
+        return metric
 
     def bestModel_evaluateImageSpacings(self, validationTransforms: Compose):
         self.network.model.load_state_dict(
@@ -437,3 +452,6 @@ def bestModel_evaluateImageSpacings(self, validationTransforms: Compose):
                 transforms.f_labelAsDiscreted(),
             ]
         )
+        self.slidingWindowInference_do(
+            self.validationSpace, self.diceMetric_onValidationSpacing
+        )