Use Case: Fully Automated 3D Segmentation and Quantification |of Liver Tumors

The reliable detection of liver tumors in CT scans and their precise measurement form the basis for effective diagnosis, surgery planning, and therapy control in liver cancer. Since manual measurement of 3D structures is extremely time-consuming, cost-intensive, and subjective, automated methods offer great promise in today’s challenging clinical environment. However, due to the many ways in which the liver can be damaged – such as alcoholic cirrhosis – and the significant variability in the appearance and shape of liver tumors, producing a reliable automated segmentation is not an easy task. Further CT image analysis difficulties can arise due to insufficient separation from adjacent organs such as kidney, heart, and muscles, and also due to time- and liver state-dependent effects of the contrast agent. Because of this, there has been a growing interest in the development of fast and accurate segmentation methods.

Researchers have utilized Definiens Developer XD to create a solution for the fully automatic segmentation of liver tumors using contextual information. Emulating a radiologist’s approach, this solution starts by identifying the liver and its adjacent structures, including lungs, spine, ribs, and gallbladder. This step alone leads to an improved separation of liver tumors from other organs. Additionally, the liver itself provides valuable local context information, which can be used to compensate any time- or liver state-dependent effects of the contrast agent. In the end, this leads to more reliable and robust identification of the tumor outlines. For example, Figure 1 shows a tumor where the information about the average intensity of the surrounding liver was crucial to identifying acceptable tumor outlines. In cases such as this, even interactive methods often fail due to the relatively dark core of the tumor.