Researchers have developed a computer-based tool to improve the accuracy of KI-67 tests, an approach that studies proteins found in cells to determine how fast they grow, an important marker used to evaluate cancer growth and progression . Leaded by Dr. Sahil Saraf, a team of scientists at Qriven PTE. Ltd. and Singapore General Hospital have collaborated on this advancement. Their research has been published in the peer-reviewed journal Heliyon.
Measuring KI-67 can help doctors understand how quickly cancer cells reproduce. However, manually examining KI-67 stained slides is time-consuming and may lead to inconsistencies among pathologists, medical experts who study tissues and diagnose diseases. To improve this, the researchers designed a computer-driven system that automates the scoring process, making the results more reliable and reducing the chance of human error. It is observed that when AI and pathologists work together, the best results can be seen. In their study, Dr. Saraf and colleagues analyzed many sarcoma cases, a type of cancer that develops in bone and soft tissue, reviewing hundreds of tissue areas under the microscope. Doctors evaluated KI-67 levels with or without computer assistance, indicating that the tool greatly improved consistency between assessments.
The results show that using a computer-based system can significantly reduce the differences between pathologists’ assessments. “The tool improves the accuracy and reliability of KI-67 readings by reducing individual interpretation differences and standardizing scoring methods,” explains Dr. Saraf. Since KI-67 is a nuclear stain, the system isolates and recognizes nuclei and It is isolated and identified and classified by advanced image processing techniques.
Doctors who initially evaluated KI-67 manually found that computer-supported methods resulted in more uniform results. “The results show that computer aid can provide basic support for pathology, helping to ensure more precise tumor classification and better treatment decisions,” Dr. Saraf noted. The study also found that doctors agree to computers in the vast majority of cases The results show a close integration between humans and automated assessments.
New technologies will change cancer diagnosis. By increasing consistency and efficiency, the tool can help provide more accurate tumor analysis, detailed analysis of tumor characteristics guided treatment, thus providing patients with better treatment options. Although human comments remain vital, they prove to be computer-assisted assessments are a valuable addition to medical practice. Further improvements in the system are expected to enhance its ability to tumor areas with higher accuracy, thereby improving its diagnostic performance.
The discovery of Dr. Saraf and his team highlights how technology can help improve medical testing, providing pathologists with reliable tools. With the increasing adoption of digital solutions in the healthcare field, this innovation will become essential for modern cancer diagnosis, resulting in better and more effective patient care.
Journal Reference
Sahil Ajit Saraf, Aahan Singh, Wai Po Kevin Teng, Sencer Karakaya, M. Logaswari, Kaveh Taghipour, Rajasa Jialdasani, Li Yan Khor, Kiat Hon Lim, Sathiyamoorthy Selvarajan, Vani Ravikumar, Md Ali Osama, Priti Chatterjee, Santosh KV. “ Improve the accuracy of reporting KI-67 IHCs by using AI tools.” Heliyon, 2024. Doi: https://doi.org/10.1016/j.heliyon.2024.e40193
About the Author
Dr. Sahil Saraf is an experienced pathologist with over 13 years of expertise in the field. He is known for his leadership, clinical skills, consulting skills, and playing a key role in advancing diagnostic techniques and research. Dr. Saraf completed a histopathology fellowship at Singapore General Hospital and held important positions, including currently known as Senior Expert, as the Medical Director of the Department of Pathology at VG SARAF Memorial Hospital, Qrisity PTE. Ltd and consultant pathologist at DDRC SRL Laboratory. In addition, he is a director of two companies specializing in chemical manufacturing.
Dr. Saraf is an important contributor to AI integration in healthcare and has developed diagnostic models for prostate grading and lymph node diagnosis. At Qriven, he played a role in getting basic certifications and helping the company gain a recognized prestigious accelerator program. Dr. Saraf’s work has been published in various medical journals and has been invited to serve as a keynote speaker at medical conferences.
In 2024, Dr. Saraf was selected as a platform presentation at the USCAP annual conference in Baltimore, Maryland, where he delivered the paper, “A deep learning module that helps pathologists identify lymph node metastasis during breast tumor resection.” He The research won him the first place in the research category of ISBP-BCRF in Larry Norton’s MD research category.
Dr. Saraf completed his MD degree at the Vydehi Center for Medical Sciences and Research and MBBS at the JJM School of Medicine. He also hired several scholarships, internships and continuing education programs.
Dr. Pritty He is a wife of Harding Medical College in New Delhi, India and a professor in the Department of Pathology. She received her undergraduate training at the Kolkata Medical College in India and conducted her M.D. pathology from the All India Institute of Medical Sciences in New Delhi, India. Her areas of interest are surgical pathology, tumor pathology, and molecular pathology. She is a passionate academician who loves teaching undergraduate and graduate students. She introduced many scientific posters and oral newspapers and won national awards for this. She has written numerous research publications in national and international journals.
Santosh KV is a consultant pathologist with 25 years of experience in diagnostic histopathology. He is particularly interested in gastrointestinal pathology and dermatology. He is also a professor of pathology and is passionate about teaching. Currently, he is involved in the AI of pathology during his spare time away from diagnostic pathology.
