Conclusions

This chapter discussed aspects related to the segmentation of medical images for the purpose of tumor evaluation and treatment assessment. Pancreatic cancer imaging by CT was used as the basis for discussing image segmentation issues for medical imaging and CAD applications. It was also used in an effort to open the pancreatic cancer imaging area into possibly more research and discussions considering that it is relatively under-investigated and unknown despite its significant toll on health care.

The current state-of-the-art in CAD methodologies for CT and pancreatic cancer was reviewed and limitations were discussed that led to the development of a novel, fuzzy logic-based algorithm for the clustering and classification of pancreatic tumors on helical CT scans. This algorithm was presented here and its pilot application on selected CT images of patients with pancreatic tumors was used as the basis to discuss issues associated with tumor segmentation and validation of the results.

The problems and difficulties encountered today by the radiologists and the oncologists dealing with pancreatic carcinoma are numerous and they are often associated with the limitations of the current imaging modalities, the observer biases, and the inter- and intraobserver variability. Among the most striking weaknesses is the inability to detect small tumors, to consistently differentiate between pancreatic tumors and benign conditions of the pancreas putting the patient through several imaging procedures and medical tests, to accurately measure tumor size and treatment effects.

Computer tools could play a diverse role in pancreatic cancer imaging. The primary goal of the system presented here was the automated segmentation of the normal and abnormal pancreas and associated pancreatic tumors from CT images. However, these tools could have a broader and more diverse role in the detection, diagnosis, and management of this disease that could change the current standard of care. Among other applications, CAD methodologies could provide objective measures of pancreatic tumor size and response to therapy that will allow (a) accurate and timely assessment of tumor resectability, (b) accurate and timely estimates of tumor size as a function of time and treatment, and (c) standardized evaluation and interpretation of tumor size and response to treatment. CAD techniques could further lead to 3-D reconstructions of the pancreas and tumors and impact surgery and radiation treatment.

Validation is and should be a major part of CAD development and implementation. Medical imaging applications, however, present unique problems to CAD validation, e.g., lack of an absolute gold standard, lack of standardized statistical analysis and evaluation criteria, time-consuming and costly database generation procedures, and other. Yet, CAD researchers are asked to find ways to overcome limitations and properly validate medical CAD algorithms including those that involve segmentation or clustering. Several options have been proposed in this chapter for this purpose. As we learn more about this area, however, we find that it may be possible to define a new family of validation criteria better suited for medical imaging applications. These criteria are likely to link algorithm performance to actual clinical outcomes. We could use, for example, classification results as a measure of segmentation performance.

10 Ways To Fight Off Cancer

10 Ways To Fight Off Cancer

Learning About 10 Ways Fight Off Cancer Can Have Amazing Benefits For Your Life The Best Tips On How To Keep This Killer At Bay Discovering that you or a loved one has cancer can be utterly terrifying. All the same, once you comprehend the causes of cancer and learn how to reverse those causes, you or your loved one may have more than a fighting chance of beating out cancer.

Get My Free Ebook


Post a comment