Annotation of lung-screening images and 2D-E proteomic analysis for early diagnosis of lung cancer through federated biobanks

Motivation and Aim: Lung cancer is the leading cause of death by tumor in Europe. New possibilities for an early diagnosis of lung cancer are provided by means of Computerized Tomography (CT) as a diagnostic device for small lung nodules early detection; so computerized techniques for the automated analysis of these scans are quickly becoming a practical necessity and are expected to provide valuable assistance to radiologists. Because the goal of an automated lung nodule detection method is to improve human detection performance, an essential component of any automated method is the process through which computer-generated results and information are conveyed to the radiologist: this improvement is possible because of expert radiologists can annotate the scans, using an automated tool and a common annotation protocol, so the results of the radiological examination and the CAD evidences can be compared. Moreover, proteomic analysis studies on plasma were aimed at the early detection of lung cancer through the identification of protein profiles and molecules involved in the pathogenesis, progression and sensitivity to specific therapies. Methods and Algorithms: Medical Application on a Grid Infrastructure Connection (MAGIC-5) is an italian academic project, involving physicists and physicians, engaged in the effort to develop CAD methods for detection of smaller nodules in screening programs. Focus of our study is the development of a prototype of working environment that can improve flexibility and efficiency in creating, storing, and exchanging annotations of medical images, associated with protein images produced by two-dimensional electrophoresis (2-DE). Therefore, we outline the most important features of the annotation protocol, that comprises both the inclusion criteria in collecting images and the clinical information consistent with our aims. Results: We describe the technical features of the low-cost automatic annotation tools, developed under Open Source platform, used by radiologists and biologists to annotate scans and protein images. Finally, we make an attempt to relate the annotations of the CT database with the results of the analysis of 2DE images, in order to identify potential pathological biomarkers by same patients. Conclusion: Our aim is to explore if proteomic analysis of the plasma could increase the sensibility and diagnostic specificity of CT analysis, also through a federation of heterogeneous biobanks of lung-screening and 2-DE images.