Doctors use the same parts of the brain to name objects and make diagnoses

A study conducted at FMUSP and University College London showed that uncovering the cerebral mechanisms of physicians can contribute to improved diagnostic methods and techniques, resulting in fewer medical errors

By Fábio de Castro

Agência FAPESP – When making diagnoses, doctors use the same cerebral mechanisms used to name objects, finds a study conducted by a group of scientists at the Universidade de São Paulo Medical School (FMUSP) and at University College London in England. The results of the study were published in the journal PLoS One.

According to the authors, their study is the first to investigate the cerebral mechanisms used by doctors to make diagnoses. The authors state that uncovering these mechanisms could contribute to improved diagnostic methods and techniques and could reduce errors. The study was carried out under the FAPESP-funded Inter-institutional Cooperation for Support of Brain Research (CInAPCe) program.

The participants in the study were Marcio Melo and Daniel Scarpin from the FMUSP Medical Informatics Laboratory, Edson Amaro Jr., Rodrigo Passos and João Sato from the FMUSP Radiology Institute and Karl Friston and Cathy Price from the Centre for Neuroimaging at UCL. Amaro Jr. is one of CInAPCe’s principal researchers.

According to Melo, the study’s main author, the results of the research show that the areas of the brain activated during the task of making diagnoses are very similar to the areas activated when naming objects. The study shows that diagnoses made from fast visual recognition of clinical signs could be based on neural mechanisms similar to those used to name objects in everyday life.

“The results of this study imply that the vast knowledge accumulated in cognitive neuroscience studies on the recognition and naming of objects could be used for improving diagnostic practices,” Melo told Agência FAPESP.

According to Melo, doctors frequently reach diagnoses in the first moments of contact with patients, sometimes even before the sick individuals tell them about their symptoms. For example, if a doctor sees a patient with jaundice, he/she immediately considers liver disease diagnoses.

“This type of diagnosis, which is immediate and automatic, is called pattern recognition. Our hypothesis was that to make diagnoses under these circumstances, doctors would use the same cerebral mechanisms that are mobilized when recognizing and identifying objects in everyday life,” he affirmed.

To perform the study, the scientists developed a new experimental model using the diagnosis of lesions in chest X-rays as an example of diagnosis based on visual information. An MRI machine used for studies at the Hospital das Clínicas Radiology Institute was used in the experiment.

“We used functional magnetic resonance imaging to detect cerebral activation of doctors who were inside the MRI machine and were making diagnoses of lesions in chest X-rays,” explained Melo.

For comparison, the doctors also identified and named animals from drawings interspersed with the radiological images. “While lying in the tube of the magnetic resonance machine, the doctors could observe the X-ray images through a system of mirrors. Each one observed 180 images: 60 of lesions, 60 of animals, and 60 of letters,” he said.

The images were projected for 1.5 seconds. The radiologist then had an additional 2 seconds to give his/her diagnosis. The average response time, defined as the time between the start of the X-ray presentation and the start of the spoken response, was 1.33 seconds. For naming the animals, the average response time was 1.23 seconds.

“These response times are an indication that this type of diagnosis can be very fast. Most of the radiologists said that in addition to the name of the lesion, differential diagnoses of some lesions came up as well. In other words, the 3.5 seconds spent with each image were enough for the doctor to think of specific diagnoses,” said Melo.

Vision and diagnosis

The results showed that the areas of the brain active during diagnosis of the lesions were very similar to those detected during the naming of animals, according to Melo.

 “The results are coherent with our hypothesis that diagnoses made from quick visual recognition of clinical signals are based on neural mechanisms similar to those used to name objects in everyday life,” he affirmed.

According to Melo, the results show that the ample knowledge accumulated in cognitive neuroscience studies on the identification and naming of objects can potentially be applied to the development of methods for increasing the competency of doctors in diagnostic tasks involving vision.

 “Aside from this, the conceptual hypothesis and the new methodological approach used in the study will be able to open new pathways to greater knowledge about cerebral processes leading to medical diagnosis,” he said.

The article “How Doctors Generate Diagnostic Hypotheses: A Study of Radiological Diagnosis with Functional Magnetic Resonance Imaging,” by Marcio Melo and others, can be read at www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0028752.