On average, doctors are 81% accurate at spotting time-sensitive secondary cells, which ideally need to be found before they break-away from their original tumour site and begin to grow in other parts of the body. Currently, a microscopic examination of a tumour in patients is considered the gold standard for cancer diagnosis but can be very time consuming and leaves room for human error.
Google ’s latest breast cancer tackling offering is called the Lymph Node Assistant (LYNA) and has been taught to recognise characteristics of tumours by studying cancer patient scans. During a trial of the software, LYNA was said to help doctors inspect scans in half the usual time and with 99% accuracy.
Nodal metastasis can occur as a result of many different types of cancer but influences treatment decisions in particular when it comes to breast cancer. Detecting the spread of breast cancer to patient’s lymph nodes can be the difference between whether they receive radiation therapy, chemotherapy, or the potential surgical removal of additional lymph nodes. This is why accuracy and timeliness of identifying nodal metastases is so important. However, Google said it has found several studies which showed that about one in four metastatic lymph node staging classifications would be changed upon second pathologic review. They also highlighted that the detection sensitivity of small metastases on individual slides can be as low as 38% when pathologists have to review them under time pressure.
Google is not alone in its quest to improve breast cancer detection methods. Recently, several other medical devices have been created to try and make lives easier for both patients and practitioners, including new imaging techniques, biosensors and blood tests.
From blood tests to biosensors
Blood tests are not often considering the most high-tech of medical devices but a study by The Institute of Cancer Research (ICR ) and The Royal Marsden NHS Foundation Trust in the UK found that a new blood test can quickly predict how well a patient can respond to the breast cancer drug, palbociclib. The test is thought to be better than existing methods because it can deliver results in two to three weeks rather than two to three months. Palbociclib was approved last year and is intended for use in patients who have not had any previous breast cancer treatments.
Researchers funded by the US National Institute of Biomedical Imaging and Bioengineering (NIBIB) had a slightly more high-tech approach when they created a new biosensor to diagnose breast cancer less invasively, compared to the existing needle biopsy approach. The biosensor chip was created by researchers from the Universities of Hartford and Connecticut and has been designed to identify the breast cancer biomarker HER-2 in a patient’s blood, all within 15 minutes. The biosensor consists of a microfluidic device and a biosensor chip coated with antibodies, which can capture and immobilise any HER-2 proteins present in a patient’s blood sample. Abnormal levels of HER-2 can indicate a specific type of breast cancer, and early detection can enable better treatment strategies.
Similarly to Google, a team of researchers at the German Cancer Research Centre (DKFZ) decided to work on an imaging-based technique to improve breast cancer detection. Their technique combines an advanced method of diffusion-weighted magnetic resonance imaging (MRI) with intelligent computer-based image analysis methods. It could help many patients avoid unnecessary control biopsies, following suspicious findings from mammography screenings, and also aid the detection of malignant changes in tissues. A study conducted using this method enabled scientists to identify malignant cases with a 98% accuracy.
Impact on patients and practitioners
The benefits of pathologists using algorithms to aid their practice and the proof-of-concept for a system like LYNA were recently explored in two research papers. The first being ‘Artificial Intelligence Based Breast Cancer Nodal Metastasis Detection: Insights into the Black Box for Pathologists’, published in the Archives of Pathology and Laboratory Medicine and ‘Impact of Deep Learning Assistance on the Histopathologic Review of Lymph Nodes for Metastatic Breast Cancer’ published in the American Journal of Surgical Pathology, which both found LYNA to be a useful pathology tool.
Google believes that its LYNA technology has the potential to reduce the burden on pathologists to conduct repetitive identification tasks and instead could allow them to spend more time and energy on more challenging clinical and diagnostic tasks. This should have a knock on positive impact on patients who could then expect more accurate and quick results which leads to quicker treatment times.
Technology in this field may be advancing rapidly but there is still some way to go before it can be considered perfect. Google thinks that LYNA’s impact on real clinical workflows and patient outcomes still needs to be observed but remains optimistic that deep learning technologies and well-designed clinical tools will improve pathologic diagnosis across the globe.