Merge pull request #178 from BrainLesion/157

added some features while doing #157

panoptica/panoptica_aggregator.py

@@ -59,8 +59,9 @@ def __init__(
         self.__panoptica_evaluator = panoptica_evaluator
         self.__class_group_names = panoptica_evaluator.segmentation_class_groups_names
         self.__evaluation_metrics = panoptica_evaluator.resulting_metric_keys
+        self.__log_times = log_times
 
-        if log_times:
+        if log_times and COMPUTATION_TIME_KEY not in self.__evaluation_metrics:
             self.__evaluation_metrics.append(COMPUTATION_TIME_KEY)
 
         if isinstance(output_file, str):
@@ -176,6 +177,7 @@ def evaluate(
             result_all=True,
             verbose=False,
             log_times=False,
+            save_group_times=self.__log_times,
         )
 
         # Add to file
@@ -208,6 +210,10 @@ def _save_one_subject(self, subject_name, result_grouped):
     def panoptica_evaluator(self):
         return self.__panoptica_evaluator
 
+    @property
+    def evaluation_metrics(self):
+        return self.__evaluation_metrics
+
 
 def _read_first_row(file: str | Path):
     """Reads the first row of a TSV file.

panoptica/panoptica_evaluator.py

@@ -193,7 +193,7 @@ def _evaluate_group(
         save_group_times: bool = False,
     ) -> PanopticaResult:
         assert isinstance(label_group, LabelGroup)
-        if self.__save_group_times:
+        if self.__save_group_times or save_group_times:
             start_time = perf_counter()
 
         prediction_arr_grouped = label_group(processing_pair.prediction_arr)
@@ -225,7 +225,7 @@ def _evaluate_group(
             verbose=True if verbose is None else verbose,
             verbose_calc=self.__verbose if verbose is None else verbose,
         )
-        if save_group_times:
+        if self.__save_group_times or save_group_times:
             duration = perf_counter() - start_time
             result.computation_time = duration
         return result

panoptica/panoptica_statistics.py

@@ -15,10 +15,16 @@
 class ValueSummary:
     def __init__(self, value_list: list[float]) -> None:
         self.__value_list = value_list
-        self.__avg = float(np.average(value_list))
-        self.__std = float(np.std(value_list))
-        self.__min = min(value_list)
-        self.__max = max(value_list)
+        if len(value_list) == 0:
+            self.__avg = np.nan
+            self.__std = np.nan
+            self.__min = np.nan
+            self.__max = np.nan
+        else:
+            self.__avg = float(np.average(value_list))
+            self.__std = float(np.std(value_list))
+            self.__min = min(value_list)
+            self.__max = max(value_list)
 
     @property
     def values(self) -> list[float]:
@@ -40,6 +46,9 @@ def min(self) -> float:
     def max(self) -> float:
         return self.__max
 
+    def __repr__(self):
+        return str(self)
+
     def __str__(self):
         return f"[{round(self.min, 3)}, {round(self.max, 3)}], avg = {round(self.avg, 3)} +- {round(self.std, 3)}"
 
@@ -181,6 +190,30 @@ def get_one_subject(self, subjectname: str):
             for g in self.__groupnames
         }
 
+    def get_one_metric(self, metricname: str):
+        """Gets the dictionary mapping the group to the metrics specified
+
+        Args:
+            metricname (str): _description_
+
+        Returns:
+            _type_: _description_
+        """
+        self._assertmetric(metricname)
+        return {g: self.get(g, metricname) for g in self.__groupnames}
+
+    def get_one_group(self, groupname: str):
+        """Gets the dictionary mapping metric to values for ONE group
+
+        Args:
+            groupname (str): _description_
+
+        Returns:
+            _type_: _description_
+        """
+        self._assertgroup(groupname)
+        return {m: self.get(groupname, m) for m in self.__metricnames}
+
     def get_across_groups(self, metric) -> list[float]:
         """Given metric, gives list of all values (even across groups!) Treat with care!