Impact of artificial intelligence in chest lesions
Since the discovery of x-rays by Wilhelm Conrad Rontgen in 1895, the utility of these new rays has extensively adopted in medicine to diagnose, prognosticate and manage many of the maladies of man. Radiograph of the chest is extensively used all over the world. In fact, statistically, this is the most common investigation sent to the departments of radiology. However, the interpretation of a chest radiograph is very difficult and the misinterpretation is 1% in radiologist groups and more than 5% in non-radilogist groups. Routine radiographs of the chest are very common and constitute more than 50% of all conventional radiographs in general and multi-specialty hospitals.
In pre-employment, pre-admission, preoperative, general health check-up and in intensive care units, chest radiographs are very essential. A tele rontgenogram is the common technique which is taken at a 6 feet distance from the centre of the beam to the patient’s chest, in deep inspiration and this should be a standard technique. This prevents and minimises the distortion of the mediastinum and heart size. In recent years, digital radiographs are the routine practice. With portable x-ray machines, it is not practical to take at 6 feet distance. Hence, the interpreter should know the technique adopted before properly interpreting the chest findings. The technique, positioning and the processing are important for proper interpretation. Even today, with best techniques by highly qualified radiographers and interpreted by eminent radiologists, a high false positive and false negative rates are observed. There has been no special training in interpretation of the chest radiographs and no regulation on who should interpret these. Today, the academic radiologists as well as private practicing radiologists do not have time for interpretation of chest radiographs as they are busy with advanced technology. Hence, all and sundry have to interpret at odd situations and at odd times, which results in misinterpretations.
To address this challenge and partly to avoid mistakes, an imaging artificial intelligence (AI) algorithm can be used so that wrong interpretations are avoided. The radiologist gets full confidence with the help of AI. The Manipal Asia radiology group and the imaging AI software developer Qure. AI has developed a deep learning algorithm arrived at the diagnosis by observing explanatory ‘heat maps’ that highlight the pathological areas. This presentation was made by Shalini Goval from Manipal Asia radiology group from Bangalore, India, at the annual conference of Radiological Society of North America (RSNA), Dec 2017 at Chicago, USA. Heat maps and ancillary tools reveal the logistics of the thinking of these algorithms which lead to proper interpretation.
This article was published Journal of Medical and Science research: https://www.jmsronline.com/article.aspx?ID=Impact-of-artificial-intelligence-in-chest-lesions
