Artificial Intelligence is helping lung cancer detection 

Mankind is facing a critical health problem: in the coming decades, cancer will become the world’s most important disease. 2023 is expected to be the year in which cancer will overcome all cardiovascular diseases, becoming the number one killer, with a tendency to double the number of cases and deaths in 2030.  

Currently, cancer kills more people than AIDS, tuberculosis and malaria combined, being responsible for 7.7 million deaths in 2007, of which 72% in developing countries. It continues to be a global problem, accounting for about 12.5% ​​of deaths worldwide in the past years, being one of the most expensive diseases to manage in terms of detection, prevention, diagnosis, treatment and follow-up.

Lung cancer is the leading cause of neoplasia deaths worldwide, with a high mortality rate. 80% of cases are related to exposure to tobacco, smoking or other forms, but also exposure to asbestos, radon and exhaust of diesel engines in crowded cities. This pulmonary disease is difficult to detect because it is hidden in the early stages, being diagnosed only in advanced stages in 2/3 of the patients. For them, the 5-year survival rate was only 4% until the introduction of immunotherapy. The rate is now closer to 20%, but early diagnosis remains just as important because many patients decide to stay home, waiting for the disease to “pass”.

The Shortfalls of Early Cancer Detection 

In the current medical practice, the diagnostic circuit takes too long, with a lot of wasted time and resources that could be better managed. In many cases, it takes at least half a year for the first symptoms to appear. A useful solution would be multiple screening programs to identify the target population, which could lead to a shortening of time until the first diagnosis. However, this requires a team of specialist doctors to be assembled and confirm the diagnosis based on the analysis for each potential case of cancer. The current shortage of doctors in the medical system, together with the lack of time needed to examine medical results is still a major impediment for detecting cancer in its earliest forms.  

The current trend in the medical field is to focus on preventive measures, starting with the linear development model of the cancer, which means that an intervention at an early stage can reduce the mortality, morbidity and the costs of treatment. According to the phases of the natural history of cancer, three levels of prevention are identified: primary, secondary and tertiary:

• The purpose of primary prevention is to reduce the frequency of cancers by identifying and possibly eliminating risk factors such as lifestyle, nutrition, stress, the environment, occupational, hormonal or genetic factors.  
• Secondary prevention involves the identification and treatment of precancerous lesions or cancers in the early stages, without clinical expression, the eradication of which can suppress the evolution towards invasive and metastatic neoplasia. Secondary prevention detects the disease after its onset and includes all medical imaging techniques and biopsies.  
• Finally, tertiary prevention is the prevention of disease recurrence after initial therapy with curative intent. 

As we can see, secondary prevention is the best way to “treat” pulmonary cancer. At this stage we are not subjected to the extraordinary variability of factors that must be taken into account from the primary prevention and at the same time we have time on our side, which is not found in tertiary prevention, when the disease has already left its mark on the human body. 

 

This is where radiology intervenes

Recent advances in medical imaging technology, such as CT and MRI scanners, have enabled the creation of increasingly detailed 3D scans. To aid in the diagnosis of the patient, these impressive images necessitate a large amount of digital data. 

Artificial intelligence (AI) will be critical in assisting with the interpretation of this medical imaging data, but it will only be feasible if healthcare professionals and AI collaborate to embrace deep thinking platforms like automated disease detection in patients. 

Computing technologies will increasingly assist doctors in making medical decisions as the amount of data they collect grows tremendously. While a physician can keep a few dozen study results and papers in mind, an AI algorithm can process more than 200 million pages in just a few seconds. Medical imaging is one of the most sophisticated areas of AI application, with exceptional sensitivity and accuracy in detecting imaging abnormalities. 

AI-assisted image processing can speed up medical scan evaluation and identification by automating searches through vast databases. By prioritizing and notifying the specific cases with cancer potential, clinicians and radiologists can analyze information faster for the patient in need. As a result, higher early detection, diagnostic performance, and prognostic value may be achieved while laboratory testing is reduced.

  

Artificial intelligence – the future of diagnostics 

This will lead in time to lower costs for performing a series of imaging scans which will facilitate early diagnosis for a large scale of people in the general population. One of the biggest benefits of medical AI assisted image processing is that it can consistently handle tedious and repetitive activities, freeing up radiologists to concentrate on more important duties and focus better towards their patients. 

While intuition and expertise are still important in a radiologist’s day-to-day work, AI may add a layer of accuracy and predictability to the search for clinical irregularities that go unreported. The decrease of radiation exposure and intravenous contrast chemicals is also a particularly pertinent to AI. This is especially important for the general population, because limiting the use of radiation reduces the risk of further chance of developing cancer. Deep learning is increasingly being utilized to improve and accelerate image reconstruction and post-processing and because of that with present technology, good image quality can be achieved with a small concentration or even no exposure to radiation. 

AI algorithms may be designed, trained, and verified using well-curated medical imaging data to predict disease activity and progress. In an urgent situation of increased prevalence of pulmonary cases in the general public, these evidence-based projections could help hospitals optimize operations. They would give reliable, quantitative data that would allow medical workers to accurately assess the severity of a patient’s sickness, allowing them to successfully triage patients and thereby ease the ever-growing patient backlog. 

As a result, using AI-powered imaging in resource-constrained settings remains a major technological and policy challenge that must be addressed as soon as possible because of its potential benefits in strengthening public health systems’ capacity to deal with this ongoing global medical problem. 

You can also read about how XVision obtained € 1 million in a new round of funding by clicking here.