Artificial Intelligence and detection of lung cancer: a better diagnostic tool


Many people face the troubles and fear associated with the test for medical diagnosis of an ailment or disease like lung cancer. However, a breakthrough in artificial intelligence is expected to relieve almost 60,000 people from unnecessary diagnostic tests for lung cancer. 

According to researchers, modern technology eliminates unnecessary biopsies that, under normal circumstances, a person has to undergo about 3300 times a year. AI promises early diagnosis of lung cancer, thus saving millions of pounds spent by the NHS annually. 

The innovative technology utilizes a computer for comparing images, thus identifying the traits of interest on the scans. This technology is based on a program developed through nodule development analysis in over 1000 patients. 

Hypothetically, from 180,000 people, cancerous lung nodules are picked per year in the UK. These nodules are used to diagnose and treat diseases like trauma followed by an accident and heart palpitations. Majority of the nodules reported are benign, i.e., it does not spread and damage other body organs. Followed by early diagnosis, the patient must undergo either one or two CT scans over three to twelve months. After that, it might be required to check the cancer growth through a biopsy. The patients who don't have cancer might be subjected to anxiety followed by unnecessary tests. 

What is happening, and how can technology help in the current scenario? 

During scans for heart and chest damage in the UK, around 200,000 cancerous lung nodules are being picked up in patients. However, most of them are harmless. But the differentiation of a nodule is based on just a scan or one image. Therefore, it requires to be monitored regularly. Notably, two CT scans for diagnosis of Lung cancer over 12 months are required. Then, a biopsy might follow it. In most cases, where the patient does not develop the tumor, it can be tiresome and an anxious process. 

With the advancement in technology and its incorporation into the healthcare industry, diagnostic methods are changing. One such example is the integration of AI technology for the early diagnosis of lung cancer. A recent study has highlighted the accuracy and efficiency of this method. The researchers have developed a program that can accurately distinguish the infected nodule. It can help to identify the future risks and thus improve the survival rates through early diagnosis. Though in high-risk patients, more CT scans and even biopsies would be required. But for patients with mild symptoms, early diagnosis and care provided with an AI program are promising. 

Optellum and AI program for detection of lung cancer

Optellum is a company rolled out of Oxford University. It has developed an AI program that can quickly and accurately distinguish high-risk nodules. It differentiates them based on which nodules require further investigation and which one could be neglected. Although it doesn't eliminate the requirement of necessary CT scans, it works by limiting their number by less than 120,00 a year. This estimate is built conservatively, as per the head of the clinical trial of the device.

According to Optellum chief executive Václav Potesil, this is the world-first technology to revolutionize lung cancer diagnosis. It has the potential to benefit tens of thousands of people each year. Doctors at the NHS have helped develop the AI program, and now the AI will serve them in the best possible way. 

The AI program of Optellum can accelerate the diagnosis whereas limiting the number of possible scans required to detect the state and stage of cancer. Moreover, it would determine the necessity of biopsy in patients who visit hospitals for cancer check-ups.

AI technology and Lung cancer detection

Current investigations and studies on the development of AI programs based on previous data of patients have helped the researchers devise a technology that is better adapted for the detection of lung cancer. 

According to researchers, the utilization of AI technology to diagnose lung cancer can deliver quick and efficient diagnosis at early stages. The research has been published in European Respiratory Society International. 

At present, computerized tomography scans are used to scan lung tumors. The scans are followed by biopsy or surgery to ensure the state of cancer, i.e., whether it is benign or malignant. For scanning, the services of an expert radiologist are required who scans and examines around 300 images to detect cancer that could be very small. 

Trials involving the usage of CT scans for the detection of lung cancer have shown promising results. However, practical limitations associated with scanning are also reported. Specifically, in this scenario, the radiologist is required to review individual scans to decide which patient requires further testing and diagnosis. 

How does this technology work? 

An AI program analyzes every radiology and CT scan report. It updates the dashboard with the newest medical data of the patient regarding any suspicious lung nodule. Thus, effectively tracking their health. Further, nodules are analyzed to allocate an AI score given the Lung Cancer Prediction LCP within the timeframe of seconds. 

Combined with a patient's relevant history given the scans and imaging, the medics can accurately determine which patient requires further testing and treatment. 

Efficacy of AI in cancer detection

The following study used data of around 888 patients previously examined by the radiologist. This was the requirement for training the AI program to identify suspicious growths. Afterward, a cohort of 1179 patients was tested using the AI program. Finally, the CT scans of these patients were taken during the last two years of the trial study. As a result, 177 patients were diagnosed with lung cancer by a biopsy following their final scan in the trial phase. 

Using AI technology, the program detected a malignant tumor in 172 patients out of 177 earlier detected with lung cancer. The results showed 97% effectiveness of AI technology in the detection of lung cancer. The missed tumors that the program was unable to detect were located near the center of the chest. This is because the tumors in the center of the chest are challenging to distinguish. However, keeping the healthy part of the body in view. 

The program was also tested using the data of one-year prior CT scans of the 1179 patients. Around 152 suspicious cases were detected that were later reported with lung cancer. 

More research and development in technology are required.

Though the results shown by recent technology have promising results, but can not overlook the probability of false positives. Therefore, the program and methodology require further testing before it is used in clinical setups. 

In the words of Benoît Audelan, a researcher in the Epione project team of the Inria (France’s National Institute for Research in Digital Science and Technology) center at Université Côte d’Azur, detection of lung cancer requires CT scans. However, detection is not an easy step. Since it requires the radiologist to examine and study the scans carefully, the shortage of radiologists for lung cancer detection can be overcome using computer programs and the latest technology. Our study has shown that the technology behind the idea can detect early signs of cancer, more specifically a year before. Our study is not aimed at replacing radiologists. But our target is to provide them with a device that can detect the earliest signs of cancer development. 

The researchers aim to develop a system that is efficient enough to distinguish between malignant and non-malignant tumors. In this way, the radiologist will be aware of which patients require further testing. Thus, helping patients save their time and energy. 

In the words of Professor Joanna Chorostowska-Wynimko, if lung cancer is detected earlier, it will improve survival rates as screening will prove to be an essential step toward the target and further treatment. In addition, research has highlighted that using CT scans for screening lung cancer can help reduce deaths related to lung cancer. She added that the current study is encouraging since Ai will eliminate the necessity of reviewing scans one by one. However, AI will improve reviewing of scans, thus allowing possible early detection of cancer. But the technology must differentiate between abnormal lung tissue that is benign from the one that is probably cancer.  


In the UK, the biggest cancer killer is reported to be Lung cancer. It accounts for around 35,000 deaths per year. It provided early detection of the ailment will render approximately 57% of the infected population to survive for five years or more as to just 3% when diagnosed at the latest stage. 

The advancement in technology has the potential to address complex health issues like lung cancer. The development of the recent AI program can detect lung cancer effectively compared to routine CT scans carried out by a radiologist. Moreover, it has the potential to limit the time frame required for studying and analyzing the scans. Thereby it provides an efficient, quick, and accurate diagnosis for the patients. AI technology is equally beneficial for patients with severe disease conditions or those forced to undergo unnecessary scans even with no such disease symptoms.