Automatic Enlarged Lymph Node Detection by Volume Estimation from 3D Abdominal CT Images Based on Speed Up Robust Features and Maximum Intensity Projection

Lymph nodes parts of immune system, play important role in clinical practice, detection is challenging due to low contrast surrounding regions, irregular shapes, variable sizes, and sparsely distributed location. detected the lymph node manually, can be time-consuming, error-prone, and depend on the observer experience. I proposed a fully automatic method for 3D abdominal lymph node detection on CT scans. Detecting lymph node important cause, the main route for metastasis and deciding resection area, so it's important to identify location and structure of lymph node to make suitable surgical procedure. the proposed method consists of candidate lymph node detection procedure, and false positive reduction steps. First Maximum intensity projection (MIP) applied to produce algorithm with low computational time and to eliminate vessels, combination done between the speeded up robust features (SURF) and the blob detection with simplified features for extracting the predicted nodes, adding new calculation parameter to the system which is the volume of the LN, to add extra accuracy for the proposed method, which allow the system from verifying the candidate LN after apply projection between the three axis of the same CT bulk. the method is evaluated for pelvic LN detection on 49 volumes containing 142 LNs, yielding an 85.9%, detection rate with 0.734 false positive per volume, with an average running time of 12–35 seconds per volume.