A review of the computer-aided detection role in the diagnosis of lung cancer in CT and PET images

A review of the computer-aided detection role in the diagnosis of lung cancer in CT and PET images
Mohammad Hossein Jamshidi,^1,* Aida Karami,²
1. Department of Medical Imaging and Radiation Sciences, School of Allied Medical Sciences, Ahvaz Jundishapur University of Medical Sciences, Ahvaz, Iran.
2. Department of Medical Imaging and Radiation Sciences, School of Allied Medical Sciences, Ahvaz Jundishapur University of Medical Sciences, Ahvaz, Iran.
Introduction: PET/CT is a powerful diagnostic tool for cancer, but it has a primary disadvantage: it generates about 1000 slice images per scan. Since most cancer screening cases are normal, radiologists must identify a small number of abnormal lesions from many images without any oversight. This can be cumbersome, along with concern regarding the deterioration of diagnostic accuracy or fluctuation of results. Here, Computer Aided Detection (CAD) provides a digital output as a “second opinion” to support a radiologist’s diagnosis and assist in evaluating many images to identify lesions and arrive at the diagnosis. In this study, we focused on the automated detection of lung tumors, such as nodules, using PET/CT images.
Methods: Different terms explored in PubMed and Google Scholar databases: lung cancer, Computer-aided detection, CT, and PET. The obtained results were selected for the title and abstracts. Finally, 16 relevant papers were selected and reviewed in full text.
Results: Many modern CAD systems for lung nodules have been evaluated using sensitivity with FPs/case of ∼5.0. Similarly, the sensitivity for detecting nodules using only CT images was 67.0%, with FP/case = 5.0. By combining CT and PET detection, sensitivity increased to 83.0% with FP/case = 5.0. Therefore, the sensitive nature of our hybrid scheme was 16% greater than that of the independent detection systems using only CT images. When the nodule size increases, it will likely merge with blood vessels, lung wall, and mediastinum. These types of nodules are difficult to detect in CT images using the detection algorithm for solitary nodules. Furthermore, significant increases in uptake can be accurately detected by PET. Because the CAD system integrates the detection abilities of two different types of imaging modalities, the sensitivity of the hybrid scheme is higher than that of the independent detection systems using either CT or PET. Both CT and PET detected 27.4% of the nodules. In contrast, 40.0% and 15.8% of the nodules were detected by CT and PET alone, respectively. These results indicated that the combination of CT and PET yields equivalent results. All the nodules in the evaluation dataset were classified into three categories based on their diameter: <10 mm, 10–30 mm, and > 30 mm. observed that most nodules with diameters <10 mm were detected using CT images. These were not detectable by PET since the small nodule SUV was decreased due to the partial volume effect. On the other hand, 91.3% of the nodules with a diameter of more than 30 mm were detected using PET images. CT detection performance decreased since large nodules do not have a massive structure by fusing the mediastinum and chest wall, while PET detection was enhanced because of significantly high uptake values.
Conclusion: CT images detect solitary nodules using CNEF that were developed previously. The PET images are binarized based on the standard uptake values (SUVs) and detection of high-uptake regions. Initial candidate nodules are identified by combining CT and PET results. FPs among the leading candidates are eliminated using a rule-based classifier and three support vector machines (SVMs) with characteristic values obtained from CT and PET images. Founded that the sensitivity of the integrated results was 83.0% with FPs/case = 5.0, which are much more desirable than those obtained via independent detection methods using CT or PET. In summary, the results indicate that this novel hybrid method may be helpful in the detection of lung cancers, perhaps, particularly in mass-screening settings.
Keywords: Lung cancer, Computer-aided detection, CT, PET.