POL5803 whole 3d workflow updated (PolusAI#289)

Author: friskluft

CMakeLists.txt

@@ -234,10 +234,11 @@ set(SOURCE
 	src/nyx/roi_cache_basic.cpp
 	src/nyx/slideprops.cpp
 	src/nyx/strpat.cpp
-	src/nyx/workflow_3d.cpp
+	src/nyx/workflow_2d_segmented.cpp
+	src/nyx/workflow_2d_whole.cpp
+	src/nyx/workflow_3d_segmented.cpp
+	src/nyx/workflow_3d_whole.cpp
 	src/nyx/workflow_pythonapi.cpp
-	src/nyx/workflow_segmented.cpp
-	src/nyx/workflow_wholeslide.cpp
 )
 
 # CLI

README.md

@@ -176,6 +176,29 @@ mdir = [
 nyx.featurize_files (idir, mdir, False) # pass True to featurize intensity files as whole segments
 ```
 
+### Whole-slide and whole-volume feature extraction
+
+Nyxus provides dedicated workflows for extracting features from whole 2D slides and volumes. The workflows scale across CPU cores as controlled by constructor parameter `n_feature_calc_threads` of classes Nyxus and Nyxus3D.
+
+Example -- whole-slide featurization with 4 threads: 
+
+```
+dir = "/2d_dataset/intensity"
+import nyxus
+nyx = nyxus.Nyxus (features=["*ALL*"], n_feature_calc_threads=4)
+f = nyx.featurize_directory (intensity_dir=dir, label_dir=dir)
+```
+
+Example -- whole-volume featurization of a NIFTI dataset with 4 threads:
+
+```
+dir = "/3d_dataset"
+import nyxus
+nyx = nyxus.Nyxus3D (features=["*3D_ALL*"], n_feature_calc_threads=4)
+f = nyx.featurize_directory (intensity_dir=dir, label_dir=dir, file_pattern=".*\\.nii\\.gz")
+```
+
+
 ## Further steps
 
 For more information on all of the available options and features, check out [the documentation](#).

docs/source/featurelist.rst

@@ -717,7 +717,16 @@ Nyxus-provided features
      - volume of the triangle mesh of ROI's convex hull
    * - 3VOXEL_VOLUME
      - total of volumes of voxels of the ROI
-
+   * - 3MAJOR_AXIS_LEN
+     - measure of the largest principal component of the ROI voxel cloud
+   * - 3MINOR_AXIS_LEN
+     - measure of the 2nd largest principal component of the ROI voxel cloud
+   * - 3LEAST_AXIS_LEN
+     - measure of the smallest principal component of the ROI voxel cloud
+   * - 3ELONGATION
+     - ratio of the two largest principal components of the ROI voxel cloud
+    * - 3FLATNESS
+     - ratio of the largest and smallest principal components of the ROI voxel cloud
 
 **3D texture features:**
 

src/nyx/dirs_and_files.cpp

@@ -415,6 +415,42 @@ namespace Nyxus
 		return 0;
 	}
 
+	int read_3D_dataset_wholevolume (
+		// input:
+		const std::string& dirIntens,
+		const StringPattern& filePatt,
+		const std::string& dirOut,
+		// output:
+		std::vector <std::string>& intensFiles)
+	{
+		if (!existsOnFilesystem(dirIntens))
+		{
+			std::cout << "Error: nonexisting directory " << dirIntens << std::endl;
+			return 1;
+		}
+		if (!existsOnFilesystem(dirOut))
+		{
+			std::cout << "Error: nonexisting directory " << dirOut << std::endl;
+			return 2;
+		}
+
+		// Common case - no ad hoc intensity-label file mapping, 1-to-1 correspondence instead
+
+		// read file names into a plain vector of strings (non-wholevolume counterpart: readDirectoryFiles_3D() )
+		std::vector<std::string> fnamesOnly;
+		std::string p = filePatt.get_cached_pattern_string();
+		readDirectoryFiles_2D (dirIntens, p, intensFiles, fnamesOnly);
+
+		// Check if the dataset isn't blank
+		if (intensFiles.size() == 0)
+		{
+			std::cout << "No intensity file pairs to process, probably due to file pattern " << filePatt.get_cached_pattern_string() << std::endl;
+			return 3;
+		}
+
+		return 0;
+	}
+
 	Imgfile3D_layoutA::Imgfile3D_layoutA(const std::string& possibly_full_path)
 	{
 		auto p = fs::path(possibly_full_path);

src/nyx/dirs_and_files.h

@@ -66,6 +66,14 @@ namespace Nyxus
 		std::vector <Imgfile3D_layoutA>& intensFiles,
 		std::vector <Imgfile3D_layoutA>& labelFiles);
 
+	int read_3D_dataset_wholevolume(
+		// input:
+		const std::string& dirIntens,
+		const StringPattern& filePatt,
+		const std::string& dirOut,
+		// output:
+		std::vector <std::string>& intensFiles);
+
 	/// @brief checks if the Tiff file is tiled or not
 	/// @param filePath File name with complete path
 	bool check_tile_status(const std::string& filePath);

src/nyx/features/3d_glcm.cpp

@@ -1253,3 +1253,12 @@ void D3_GLCM_feature::reduce(size_t start, size_t end, std::vector<int>* ptrLabe
 		f.save_value(r.fvals);
 	}
 }
+
+/*static*/ void D3_GLCM_feature::extract (LR& r)
+{
+	D3_GLCM_feature f;
+	f.calculate(r);
+	f.save_value(r.fvals);
+}
+
+

src/nyx/features/3d_glcm.h

@@ -127,6 +127,7 @@ class D3_GLCM_feature : public FeatureMethod, public TextureFeature
 	void save_value(std::vector<std::vector<double>>& feature_vals);
 	static void parallel_process_1_batch(size_t start, size_t end, std::vector<int>* ptrLabels, std::unordered_map <int, LR>* ptrLabelData);
 	static void reduce(size_t start, size_t end, std::vector<int>* ptrLabels, std::unordered_map <int, LR>* ptrLabelData);
+	static void extract(LR& r);
 
 private:
 

src/nyx/features/3d_gldm.cpp

@@ -542,3 +542,9 @@ void D3_GLDM_feature::reduce (size_t start, size_t end, std::vector<int>* ptrLab
 	}
 }
 
+/*static*/ void D3_GLDM_feature::extract (LR& r)
+{
+	D3_GLDM_feature f;
+	f.calculate(r);
+	f.save_value(r.fvals);
+}

src/nyx/features/3d_gldm.h

@@ -47,6 +47,7 @@ class D3_GLDM_feature : public FeatureMethod, public TextureFeature
 	void osized_calculate(LR& r, ImageLoader& imloader);
 	void save_value(std::vector<std::vector<double>>& feature_vals);
 	static void reduce (size_t start, size_t end, std::vector<int>* ptrLabels, std::unordered_map <int, LR>* ptrLabelData);
+	static void extract (LR& r);
 
 	// 1. Small Dependence Emphasis(SDE)
 	double calc_SDE();

src/nyx/features/3d_gldzm.cpp

@@ -24,6 +24,13 @@ void D3_GLDZM_feature::reduce (size_t start, size_t end, std::vector<int>* ptrLa
 	}
 }
 
+/*static*/ void D3_GLDZM_feature::extract(LR& r)
+{
+	D3_GLDZM_feature f;
+	f.calculate(r);
+	f.save_value(r.fvals);
+}
+
 void D3_GLDZM_feature::clear_buffers()
 {
 	f_SDE =