Artificial intelligence (AI) is rapidly driving digital oncology. Digital biomarker testing can help clinicians make informed and personalized decisions in cancer treatment. However, as of 2023, there are still few such products on the market that have been maturely used on a large scale.
A consensus report involving 24 experts with first-hand computational pathology/pathology AI (CPath/AI) experience says AI will improve diagnostic accuracy and will significantly change the daily tasks of pathology technicians. By 2030, AI will be used routinely and effectively in pathology laboratories.
Two independent studies recently published in eBiomedicine and The Lancet Digital Health, a subsidiary journal of The Lancet, respectively looked at: the prognostic significance of a deep learning-based tumor-infiltrating lymphocyte (TILs, which can be used as a drug target for cancer treatment) scoring system in different stages of melanoma (a skin cancer); and the value of using AI as an independent reader in the mammography workflow.
Two studies on different cancers
The first study was jointly completed by researchers from the Department of Dermatology at the University of Tuebingen in Germany, the Department of Dermatology at the University of Heidelberg in Germany, and the Department of Pathology at Yale University School of Medicine in the United States. In the study, the researchers used the deep learning algorithm NN192, an algorithm developed for the standard and digital TILs scoring system "eTILs," to analyze 321 primary melanoma and 191 metastatic samples.
The researchers found that melanoma patients with low eTILs scores had more than twice the risk of developing distant metastases from their cancer tissue than patients with high eTILs scores. At the same time, eTILs scores decreased between primary melanoma and metastasis samples. Patients with eTILs score ≤12.2% and concurrent anti-PD-1 immunotherapy treatment have poor survival outcomes. This demonstrates that eTILs are predictive of primary melanoma samples and that eTILs can predict response and survival outcomes in patients receiving PD-1 therapy.
In this regard, Roberto Salgado, co-chairman of the International Immuno-Oncology Biomarker Working Group, said that accurate quantification of immune cells involves prognostic and predictive information and is important for clinical pathways and customized treatment plans. In addition, computer assessment results are much more accurate than manual assessment.
The second study was conducted by Karin Dembrower and her team at the Department of Oncology Pathology at the Karolinska Institute and Capio Sankt Göran Hospital in Sweden.
In this study, the research team included 55,581 women aged 40-74 years with unfilled breast implants based on regular breast cancer screening at Capio Sankt Göran Hospital from April 1, 2021 to June 9, 2022. The study followed the Swedish National Guidelines formatammography screening, in which two radiologists independently evaluated each participant's mammogram, and in the event of an abnormal reading by any one of them, a consensus discussion was held to decide whether to proceed with further imaging. If further testing still suspects the patient has cancer, a biopsy sample is obtained, which is analyzed by a pathologist and a definite diagnosis is made.
In the study, InsightMMG (an AI system) ran as an independent reader in the background while two radiologists read the images. Radiologists were unable to access InsightMMG for information prior to the consensus discussion, during which radiologists had access to InsightMMG information for all cases, including any local image findings, graphical contours, and corresponding AI abnormality scores.
The research team conducted four reading strategies and examined the actual diagnostic results of double reading by two radiologists (standard situation), double reading by a radiologist and the AI system, single reading by the AI system, and third reading by two radiologists and the AI system. The results showed that compared with the standard situation, the cancer detection rate for double reading by a radiologist and the AI system increased by 4%, and the recall rate decreased by 4%; the cancer detection rate for single reading by the AI system had no significant difference, and the recall rate decreased by 47%; the cancer detection rate by two radiologists and the AI system for third reading slightly increased, the recall rate increased by 5%, and consensus discussions increased by nearly 50%.
The research team said that AI systems and humans will regard certain different image features as suspicious cancer when reading pictures, so the synergy of humans and AI systems can improve the detection rate of breast cancer in mammograms. The AI system's single reading minimizes the psychological burden on participants caused by multiple exams, but it means that a large proportion of mammograms are never evaluated by a doctor. Two radiologists and the third reading of the AI system can detect cancer to the greatest extent, but this must be balanced against issues such as increased detection costs and a shortage of radiologists.
The market still needs to continue to develop
Roberto Salgado said that digital biomarker testing can help clinicians make informed and personalized decisions in cancer treatment. However, as of 2023, there are still few such products on the market that have been maturely used on a large scale.
On September 7, local time, Paige.AI, an American cancer diagnostic technology developer, announced that it would cooperate with the American technology company Microsoft (Microsoft) to build the world's largest image-based AI model and apply it to the development of digital pathology and oncology.
Coincidentally, on September 11, local time, the American technology company Dell (DELL) and the Digital Cancer Research Center of the University of Limerick in Ireland jointly developed an AI platform and digital twin technology to promote prediction and diagnosis research of B-cell lymphoma.
"This is a very exciting start, and we look forward to the digital support of the Dell Technologies team to accelerate this project." Paul Murray, Professor of Molecular Pathology at the University of Limerick and Scientific Director of the Digital Pathology Unit of the Digital Cancer Research Center, said, "By working with the Dell Technologies team, we will be able to further understand how cells go wrong during cancer development and find new ways to diagnose and treat cancer patients."