Pixel pathology: digital workflow and computational analysis

Abstract

In pathology laboratory it recently becomes possible to make digital images with a whole slide scanner. This provides the pathologist with the possibility to view microscopic images on a computer screen. It is pivotal that the scanned microscopic images are of good quality and resemble the images seen through the microscope as much as possible. It is expected that digitalization in pathology diagnostics will improve in quality, safety and efficiency. The Laboratory for Pathology East Netherlands (LabPON) in liaison with other partners started testing several scanners and digital pathology software to improve the workflow for digital diagnostics. This thesis describes this process. After the transition to digital diagnostics LabPON put a lot of effort into optimizing the working conditions for pathologists. It is well known that there are health issues related to working with the microscope. In principle, these ergonomic complaints can be prevented when working digitally for diagnosing tissues. On the other hand, new health issues (like repetitive strain injury) may arise for the pathologist. At LabPON we tested different types of computers and monitors as well as new user interphases were used and adapted to create an optimal work environment for the pathologist. In order to work efficiently for the pathologist the Laboratory Information and Management System (LIMS) and Image Management System (IMS) are completely integrated into the LabPON workflow. Permutations of patient information and material can be prevented by working in this manner. Another interesting feature of working with the digital method of diagnosing is that it is possible to work outside of LabPON and enables you to easily consult specialists from other pathology laboratories. We compared the quality of consultation with tissue slides and the digital method of consulting tissues. The digital consultation was more efficient than sending patient material (i.e. tissue slides) by post and the time for diagnosing a patient was reduced by 1 to several days. In addition, the risk of damaging or losing unique patient material when sending by post is eliminated. In chapter 4 we demonstrate that with the transition to digital diagnostics we saved 19 hours of lab and administrative work per day in LabPON. This makes digital diagnostics more efficient and more cost effective. With diagnosing digitally the pathologist seems to be 12% more efficient than with the microscope. With the use of special software, digital tissue images can be analysed as well. This software can improve the efficiency but also the quality of the diagnosis. An additional benefit of using software is that it can do a lot of pre-work for the pathologist. At LabPON together with our partners we demonstrated that the tumor cell and tumor stroma ratio in colon cancer can be determined more accurately when using digital pathology software. In general, digitalization of diagnostics in pathology laboratories and the use of image analysis software will help improve the quality and efficiency in pathology practice, as well as improving patient care.