Note: This research product won a silver medal at the International Trade Fair (Ideas-Inventions-New Products) held at the Nuremberg Exhibition Center, Germany from 3 to 6 November 2005.
Researchers:
Dr. Nor Ashidi Mat Isa (Contact details available to registered journalists)
Dr. Harsa Amylia Mat Sakim
Assoc.Prof. Mohd. Yusoff Mashhor
Prof. Dr. Nor Hayati Othman
Norhidayati Abu Seman
The Current Problem
1) Cancer of the breast is the most common cancer in women.
2) Annually there are 450,000 new cases every year.
3) This cancer can be screened by mammogram or breast self examination.
4) Mammogram is expensive and is only available in big.
5) Mass screening has been a problem in many developing countries because:
- Expensive equipment
- Currently there is no existing diagnostic system which could help reduce the workload of pathologists
- Feeling fatigue after about reading about 30 slides. Tiredness may lead to errors
The Solution : NeuralMammo
NeuralMammo is an artificial intelligent system to diagnose breast cancer based on images captured from fine needle aspirates. The system consists of two main components; i.e. an image processing tool and an intelligent diagnostic tool.
It detects abnormal cervical cells and classify the breast lesions into three categories, namely fibroadenoma, fibrocystic disease, malignant and other benign disease. The diagnostic criteria used are the same criteria pathologists use on diagnosing breast cytology.
It has a high accuracy up to 99.3% , high sensitivity 97.5% high specificity up to 100% meaning there is no false positivity (the specificity of conventional pap smear is only up to 86%). NeuralMammo has low false negativity and false positivity of 2.5% and 0% respectively.
NeuralMammo also provides the degree of the confidence level of analysis to indicate the confidence level for each diagnosis. This enables the borderline cases to be double-checked by pathologist.
Diagnosis can be obtained within a few seconds after the user gives command to the system therefore a larger volume of samples can be processed over a short duration of time. It does not get tired with time, therefore mass screening is possible
The Novelty Of NeuralMammo
1) Nobody has developed such a system before
2) The artificial brain behind the system which uses combined hybrid Neural network (NN) and hierachical NN — thus giving accuracy, specificity, sensitivity approaching 100 %. The conventional NN gives specificity up to only 80%
3) It has a specialized image processing procedure
4) NeuralMammo provides intensive image processing facilities that could be used by pathologist to enhance the features of the images on the slide.
5) This is particularly useful for difficult cases and hence it will help to reduce false negative and false positive rates.
6) NeuralMammo also provides automatic diagnostic function that could be used by pathologist as a second opinion.
This System Has A Global Impact
1) This is the first of its kind that is of high specificity, sensitivity and accuracy
2) FNA can be done by surgeons or other trained staff and the smears can be sent for analysis using Neuralmammo
3) NeuralMammo could do initial screening of the breast lump before they are confirmed by pathologists
