in house developed matlab toolbox antx  (MathWorks Inc)

Journal: Heliyon

Article Title: An integrated radiology-pathology machine learning classifier for outcome prediction following radical prostatectomy: Preliminary findings

doi: 10.1016/j.heliyon.2024.e29602

Figure Lengend Snippet: Visualizations of one of the topmost features of M R (orginal_collage2D_glcmV_JointEnergyEntorpy) (R1-R4) and M P (Shape: 5 %/95 % invariant 1) (P1–P4) between four different patients. The columns 1,2 represent patients with a low risk of rising PSA and columns 3,4 represent patients with a high risk of rising PSA. It can be observed that the visualizations of Co-occurrence of Local Anisotropic Gradient Orientations (CoLlAGe) gray level cooccurrence matrix (GLCM) radiomic feature on apparent diffusion coefficient (ADC) maps indicates the presence of higher density of high entropy regions for which M R has classified as rPSA + ( : R3, R4), as compared to the ones for which M R has classified as rPSA − ( : R1, R2). Similarly, the pathomic visualizations of Shape: 5 %/95 % invariant 1 depicts that a high risk of rising PSA with more aggressive cancer leads to uniformly small, malformed lumen resulting in a lower 5th/95th percentile ratio (lower range) ( : P3, P4) as compared to cases with lower risk of rising PSA ( : P1, P2). For radiomic visualizations, the feature array output from the pyradiomics package was used to overlay on top of the ADC using matplotlib package. For pathomics visualizations, in-house MATLAB code was used to overlay the visualizations.

Article Snippet: The pathomic feature extraction was performed using an in-house MATLAB V.2019 (MathWorks, Natick, Massachusetts, USA) toolbox.

Techniques: Diffusion-based Assay