Researchers:
Dr Nor Ashidi Mat Isa (Contact details available to registered journalists)
Nor Rizuan Mat Noor
Assoc. Prof. Dr Mohd Yusoff Mashor
Prof. Dr Nor Hayati Othman
Sariah Rajuli
Nazahah Mustafa
CCAID System-An Automatic Intelligent Diagnostic System for Cervical Cancer
Cervical cancer is the second most common cancer among women after breast cancer.
Mortality rate due to this disease can be reduced if mass screening program is possible. However, this aspiration cannot be realized because the problem of shortage of pathologists in Malaysia.
Previously, an intelligent software system to diagnose cervical cancer called NeuralPap has been invented. NeuralPap has the capability to diagnose cervical cell images, based on conventional Pap smear slide. However, pathologists are moving into the
latest technology in screening test for cervical cancer, which is called the ThinPrep, As a result NeuralPap is no longer the best choice for cervical cancer screening tool.
The tediousness of the NeuralPap system may now be addressed by our new invention called “CCAID System”. CCAID System is an automatic intelligent diagnostic system for cervical cancer. This invention consists of two main components. The
first component, namely an automatic and fully computerized light microscope is used to scan and capture images from the whole areas of cervical smear slides automatically under 100x and 400x magnifications. These images will be stored into the second component, namely diagnostic software.
For a typical 400x magnification, there are about 1,000 to 1,500 images; depending on the resolution of the image taken. It is tedious for pathologist to screen all of the images. To overcome this problem, this invention employs fuzzy logic concept to
automatically rank the captured images based on the most probable abnormal cases to normal cases. This ensures an easier and more systematic diagnosis process. It also imitates the actual pathologist in the capability to display the image ina 400x map and is switchable to 100x magnification for any area inside. All these images will be
displayed in thumb nails format navigation.
After the process of storing, the second component of the CCAID system will automatically extract several features (i.e. size, perimeter, colourintensity) of cervical cells. Based on these features, the CCAID system will classify the cells into 3 main cases; normal, LSIL and HSIL cases. With high Accuracy (94.29%), sensitivity (89.42%) and specificity (100%), and low false negative (5.29%) and false positive (3.94%), this system is capable of diagnosing ThinPrep images easily with high reliability.
Hierarchical hybrid neural network is used as the artificial brain that is specially designed to optimise the diagnostic performance. To improve the visibility of cervical smear images, CCAID System provides image processing facilities such as contrast enhancement and noise filtering, for ease in automatic segmentation and features
extraction. Level of confidence is also available as an indicator for the confidence of each diagnosis, simulating the normal human thinking! As a result, the low level of confidence can be considered as borderline case and need to be further diagnosedby pathologists.
There are a lot of advantages of this CCAID system. It has the potential to tremendously reduce the pathologist workload and at the sametime gives good performance. The software component comes with a user friendly interface that makes it easy to use even after a short period of training. Even a cytotechnologist can take
advantage of this system. Therefore, this invention can be used as a mass screening tool for cervical cancer. Each and every cervical smear test will be tested using this invention and only the suspected positive cases will be passed to pathologists for
further examination.
Therefore, this CCAID System, the first automatic diagnostic system for cervical cancer ever invented in the world, has great potential to be commercialised as a mass
screening tool especially in developing countries, including Malaysia.
Note: The CCAID System won a silver medal at the recent Malaysian Technology Expo
