First 2023 Research brown bag.

During the last Friday of every month, lunchtime is always busy as preparations for the Research brown bag sessions are underway. The plausible discussions, some of which continue in the corridors after the Research brown bag is over, usually set the precedent for meaningful engagements and future sessions. During Research brown bag sessions, researchers,(students and staff) from the University present summaries of their research ideas or findings from their research and how they are using research to look into the challenges that beset society.
The research year 2023 started in full throttle at Kifaru room where a Fourth-year finalist from the School of Computing and Engineering sciences , Baluge Akiza Innocent, graced the event with a research project on the Brain Cancer Classification model. According to the World Health Organization, early diagnosis improves cancer mitigation by providing care at the earliest possible stage and is an important public health strategy in all settings. Hence, having efficient computer-based models that can aid with diagnosis would be a huge leap forward in improving cancer patients’ treatment.
Innocent’s project aimed at building a model that not only classifies brain cells as tumorous or not but detects the location of cancer in the cells at an early stage which will be used in different medical diagnosis applications to generate accurate and faster results. “There is a need for identifying existing techniques and technologies used for cancer development, point out shortcomings of existing models then design and develop a model to classify and localize cancerous cells.” Innocent explained.
Using the deep learning model, a subset of the Machine learning(ML) algorithm, which consists of algorithms that are inspired by the functioning of the human brain, he utilized the CNN architecture which minimizes human effort by not only automatically detecting the location of tumors in the cells but also classifies cells as tumors or not. The model was tested through a demo and achieved a 96% accuracy level. However, the results were only limited to the data set of 3000 MRI images. The project led to the development of a web application that can identify brain cancer from an MRI scan image.
One of the standout questions that arose from the audience was whether there was a need for the model taking into consideration that professional radiologists can visibly detect tumors from MRI scans. “We can’t understate the fact that human error is inevitable and with matters concerning human health, taking chances shouldn’t be an option either. There is a need to have developed and approved algorithms that assist professionals in detecting early Brain cancer tumors. The collaborative effort is imperative in achieving proper results.” Innocent responded.
“In the future, we can leverage the need to train our model’s ability to classify other types of cancers other than brain tumors since brain cancer has different classes.” Innocent added. Developing proactive cancer-detecting models is the future of effective treatment with fewer side effects and improved long-term survival. Progress will rely on a detailed understanding of upcoming models and a robust evaluation of the implications for individuals and society.
The article was written by Kevin Mwangi.
In case you have research projects or ideas for presentations, you can register using the link here.
The Brown Bag sessions are conducted physically every last Friday of the month.