Tumor Sensitive Matching Flow: An Approach for Ovarian Cancer Metastasis Detection and Segmentation

Thursday, October 11, 2012 — Poster Session III
10:00 a.m. – Noon	Natcher Conference Center, Building 45	CC	IMAG-16

Authors

• JF Liu
• SJ Wang
• MG Linguraru
• RM Summers

Abstract

Accurately detecting and segmenting ovarian cancer metastases have great clinical impact on diagnosis and treatment. We propose a tumor sensitive matching flow (TSMF) to identify shape variance caused by metastasis between patient organs and healthy organs. Instead of separating tumor feature computation and classification, TSMF juxtaposes two rules within a partial-differential-equation framework. Metastases can be accurately located by choosing areas with large TSMF vectors, and segmented by exploiting the level set algorithm on these regions. The proposed algorithm was validated on contrast-enhanced CT data from 11 patients (age range, 25-64 years; mean age 49.3 years ± 12.9 [standard deviation]). The size range of 26 metastases is 4.0-49.9mm (mean size 18.9mm±11.9). 84.6% (22/26) of metastases were successfully detected. Two metastases less than 5mm were missed, and the other two were missed due to incomplete organ segmentation. The average false positive rate per patient was 1.2. The segmentation error of all detected metastases was 3.8±0.9mm by measuring average distance between the segmentation results and the ground-truth. The validation results demonstrated that the TSMF algorithm accurately detects and segments metastases larger than 5mm.

back to top