Presentation on theme: "Tumor Discrimination Using Textures"— Presentation transcript:
1Tumor Discrimination Using Textures Presented by: Maysam Heydari
2IntroductionMain goal: Discrimination between different tumor grades/types using textural propertiesTumor pathologies:Grade 2: astro (7), oligo (22)Grade 3: aa (2), ao (1), amoa (1)Grade 4: gbm (17)
3Introduction Patient data: 50 unique patient-study pairs: 25 expert segmented patients25 Maysam segmented patientsFor each patient, the study nearest to the biopsy date (in the range ±52 weeks) was picked.The nearest biopsy was chosen to determine the pathology.
4Weeks between study and biopsy Expert segmentedMaysam segmented(low grade tumors)# of patientsweeksweeks
5Texture FeaturesFeatures extracted on the segmented tumors: ENH (T1, T1C) and EDE (T2) on every slice.Each pixel in the tumor receives a texture intensity:Gray Level Co-occurrence Matrices (GLCM)MR8BGLAM left-to-right symmetry similarity values
6Texture Features GLCM stat measures: Energy: “orderliness” of pixels Contrast:
7Texture Features MR8 filter bank: For each filter, max response over 6 orientationsFilters:3 scales of edge filters3 scales of bar filtersA GaussianLaplacian of Gaussian
8Texture Features BGLAM: Texture similarity of the segmented tumor to the symmetric side of the brain.
17What’s Next? Combine the histograms from several texture features … Stack them as vectors?Curse of dimensionality … with only 50 data inputs.Instead of histograms, use stats: mean, var, min/max?Supervised learningSVM?