AI transforming digital pathology of cancer diagnosis
Asia is home to nearly half of the world’s population with cancer, with 9.5 million new cancer cases and 5.8 million cancer deaths reported in 2020. This can be traced to population growth, ageing population, and lifestyle or socioeconomic changes.
With a burden estimated to double by 2040 to 30.2 million cancer cases, swift prevention, detection, treatment and supportive care programmes are critical to saving lives.
Pathologists from the core of cancer care delivery for cancer detection and treatment and oncologists (doctors who treat cancer). Pathologists are specialists who examine diseased cells and tissues for the presence of cancer (i.e. malignancy) and help decide the further course of treatment. The pathologist’s investigation impacts nearly 70 per cent of clinical decisions.
Challenges in pathology
The rising cancer trend and the backlog of cases due to the COVID-19 pandemic have brought the need for trained pathologists to the fore. With advancements in R&D, more effective pathology tests are now available for cancer detection and evaluation.
However, their application requires an experienced workforce. Insufficient inputs can cause delays in cancer diagnosis, consequently impacting patients’ treatment and survival.
Key challenges in this sector include an overall shortage of pathologists, a lack of a trained workforce (it can take up to 15 years to train pathologists) and increasing complexity of analysis, all of which need to be addressed to ensure the quality of cancer detection and staging remains uncompromised.
Under these circumstances, pathologists continue to face tremendous pressure and concerns, such as the possibility of burnout.
Digital pathology facilitates remote diagnosis
Depending on the tissue being analysed, a pathologist performs tests and typically views the tissue on a glass slide under a microscope to arrive at disease-specific scores or diagnosis.
Digital pathology, in simple terms, allows scanning of the slides or whole slide imaging (WSI) so they can be viewed on a computer monitor. Thus, high-resolution images can be conveniently viewed from remote sites without travelling and can even be shared to facilitate consultation with other specialists.
While the COVID-19 pandemic accelerated digitalisation in several other sectors, its use in pathology is still nascent.
As digital adoption accelerates, the digital pathology landscape in Asia will continue to evolve. By 2027, it is estimated to reach US$125 million from US$74 million this year.
Integrating AI into pathology
AI involves computers that mimic the processes of learning and interpretation by the human mind. AI-integrated digital pathology can help streamline workflow, enhance efficiency, and improve diagnostic concordance.
On the one hand, it can help automate several complex tasks. For example, computer-based analysis of the digitised tissue images can assist with time-consuming yet essential diagnostic tasks (e.g. counting the total nuclei or classifying tumour tissue), easing the overall workload of pathologists.
Conversely, AI integration enables the system to ‘learn’ image analysis per pre-established parameters, thereby supporting the pathologist’s diagnostic acumen. For example, AI has shown to be capable of assessing features indicative of high-risk colorectal cancer from tissue images.
AI-based digital imaging analysis also helped detect metastatic areas (cancer which has spread to other organs). It showed great sensitivity in scoring certain markers that predict survival in breast cancer.
Hence, it can be used as an effective screening tool for detecting malignancy and metastases and evaluating prognosis. These assistive technologies help improve efficiency, freeing up pathologists’ time for other important tasks requiring specialised input.
When associated with pathology, deep learning (a more complex subtype of AI) can assist with diagnostic evaluation (e.g. differentiating between diseased and normal tissue, grading cancer or distinguishing cancer types) and offer deeper disease insights.
This includes predicting the status of gene mutation, outcomes and disease recurrence. The synergy of human and AI-based insights in cancer pathology no doubt opens avenues for early cancer detection, with the potential to offer patient-specific disease assessment and treatment. Ultimately, this inflicts less strain on the healthcare system and enhances patient care.
Lastly, managing pathologists’ perceptions of AI and enhancing their adoption is critical to aiding this transformation. Educating and supporting pathologists using such cutting-edge technology can help alleviate pressure on the system and manage the rising disease burden.
AI-integrated digital pathology can thus work in synergy with the pathologist to deliver optimal patient-centric care, right from detection to the treatment of cancer.