Cancers of the Lung and the Rise of Artificial Intelligence
Article Sep 26, 2018
Whether you realize or not, we are witnessing the extraordinary growth of artificial intelligence (AI) right before our eyes. While it has been introduced across countless industries, it has begun to transform the world of healthcare and researchers across the globe are now dedicated to discovering its potential. This technology has particularly struck a chord with cancer research surrounding early detection, a long-standing concern for the cancer community.
Advancing Medical Research
Although the concept of AI may seem overwhelming, it can be broken down into two simple approaches: machine learning and deep learning. Machine learning is focused on translating and understanding numerical data, while deep learning is focused on how a computer can learn to recognize key features within an image. One facet of AI that is used on a nearly daily basis is image recognition. An example of this form of deep learning include Facebook’s ability to automatically tag friends in photos, show suggested content by a streaming service based on what you’ve previously viewed, or virtual personal assistants such as Siri. Much of this technology is used frequently without a second thought, but researchers are now recognizing its groundbreaking potential for oncology.
Chen Kuan, the Founder of Infervision, a machine learning and computer vision company focused on cancer diagnostics, has dedicated his work to deep learning and image recognition to combat the leading cause of death across China: lung cancer. Kuan recognized the shortage of physicians, which often forces doctors to evaluate hundreds of thousands of images and causes exhaustion and ultimately diagnostic errors. His research enabled AI to predict if an X-ray is normal, allowing radiologists to solely focus on the abnormal images and save an immense amount of time and energy.
Researchers have even gone so far as to suggest that just 10 years from now, all medical imaging will be initially reviewed by AI before reaching radiologists. This advancement will help specialists prioritize images that require immediate action versus those with less urgent concerns, all while collecting data and recognizing key patterns that indicate cancer or other life-threatening illnesses.
AI and Radiology
Accounting for 21 percent of cancer deaths, lung cancer continues to be the leading cause of cancer-related death in the United Kingdom (U.K.). While low-dose computed tomography (CT) screening has become paramount to early detection, AI is slowly gaining momentum within diagnostics.
Just last year, an annual meeting held by the Radiological Society of North America (RSNA) deeply discussed how AI can help radiologists better detect lung cancer. The spotlight fell on Canadian researchers who developed and found success with a machine learning model that could identify a single malignant lung nodule. Dr. Aerts, a researcher at Harvard University, went on to express that AI is able to go beyond what the human eye is capable of seeing and can detect the subtle, yet critical, differences between benign and malignant tumors.
In addition, the University of Central Florida in the United States recently announced that engineers have developed an AI model that taught a computer how to identify microscopic traces of lung cancer in CT scans. Although this is a well-known challenge for radiologists, the model’s results were around 95 percent accurate, while they are closer to 65 percent accuracy when done manually. The team at UCF plans to partner with hospitals to implement this model and has been invited to present their research this month at the International Conference of Medical Image and Computing and Computer Assisted Intervention (MICCAI), the largest conference for medical imaging research thus far.
Other diseases of the lung, such as pleural mesothelioma and pulmonary hypertension, have seen some progress with the help of AI. Mesothelioma, an aggressive form of cancer that occurs as a result of asbestos exposure, continues to stump researchers in the field of oncology. This rare cancer most commonly affects the lungs and the Health and Safety Executive reports that 2,595 people died from mesothelioma in the U.K. in 2016 alone. Researchers have yet to discover a standard diagnostic procedure or form of treatment that consistently offers patients better outcomes.
Researchers in Turkey created an artificial immune system (AIS) to target malignant pleural mesothelioma patients (MPM). The 2015 study compares the results of the AIS to classify cancers against the results of the multi-layer neural network. Results showed that the system successfully classified data with 97.74 percent accuracy, as compared to the 91.3 percent accuracy by a dataset from Dicle University, Medicine Faculty Hospital based on patients’ reports. This study proves that AIS may be more efficient than previous diagnostic methods and may better aid doctors in the classification of MPM.
The rise of AI in the healthcare field has proven that a cure for cancer may exist, but there is more work yet to be done. While some see the endless possibilities, others believe it may diminish the value of real-life doctors and critical thinking. A report by STAT claimed IBM’s supercomputer arrived to invalid cancer treatment conclusions, which has raised concerns surrounding doctors failing to recognize when the system has made an error. This has prompted professionals to seek standard regulations across the ever-changing world of healthcare technology in order to ensure patients are safe and receiving the best treatment possible.
While these concerns are valid, research continues to demonstrate how impactful AI will be for detecting and diagnosing cancer earlier than ever before. Once proper regulations are put in place, AI has the potential to reduce the rate of false-positives and to advance nodule detection and evaluation in the future.
To pick apart the differences between individual cells in complex multicellular organisms, we need to look at cells one-by-one. This article takes a look at how several scientists in North America are using single cell proteomics (SCP) technologies to discern disease pathogenesis and enhance directed stem-cell differentiation.READ MORE
Healthcare interoperability remains an intractable problem; both healthcare providers and patients feel unease about limited access to medical data, which results in lower care quality and poor patient engagement. Although daunting, these challenges aren’t insurmountable. In this feature, we’ll explore what has been and what could be done to foster the creation of an interoperable infrastructure.READ MORE
5th International Congress on Epigenetics & Chromatin
Aug 22 - Aug 23, 2019
International Conference on e-Health and Healthcare Innovations
May 08 - May 09, 2019
12th International Conference on Cancer Stem Cells and Oncology Research
Jul 18 - Jul 19, 2019