Automated tumor partitioning
Abstract:
The invention provides for a medical instrument (100, 300, 400) comprising: a memory (110) for storing machine executable instructions (112) and a processor (106) for controlling the medical instrument. Execution of the machine executable instructions cause the processor to: receive (200) three dimensional medical image data (114) descriptive of a subject (318), wherein the three dimensional medical image data comprises voxels; receive (202) a segmentation of the three dimensional medical image data, wherein the segmentation divides the three dimensional image data into non-tumor voxels (700) and tumor voxels (500); choose (204) a center point (118) of the tumor voxels; divide (206) the tumor voxels into multiple groups (120) using a set of orthogonal planes (502, 504, 600), wherein the center point is within each of the orthogonal planes; calculate (208) at least one group radiomic feature (122) selected from a set of radiomic features for each of the multiple voxel groups; compute (210) a statistical measure (124) for each of the at least one group radiomic feature; calculate (212) a scalar value (128) by calculating the sum of each statistical measure multiplied by a predetermined group weighting value (126), wherein the predetermined group weighting value is unique for each statistical measure; and provide (214) a signal using a signaling interface device (108, 402) if the scalar value is above a predetermined threshold (130).
Public/Granted literature
Information query
Patent Agency Ranking
0/0