𝐁𝐫𝐚𝐢𝐧 𝐃𝐞𝐜𝐨𝐝𝐢𝐧𝐠 (𝐃𝐞𝐞𝐩 𝐈𝐦𝐚𝐠𝐞 𝐑𝐞𝐜𝐨𝐧𝐬𝐭𝐫𝐮𝐜𝐭𝐢𝐨𝐧)

 𝐓𝐨𝐩𝐢𝐜 - 𝐁𝐫𝐚𝐢𝐧 𝐃𝐞𝐜𝐨𝐝𝐢𝐧𝐠 (𝐃𝐞𝐞𝐩 𝐈𝐦𝐚𝐠𝐞 𝐑𝐞𝐜𝐨𝐧𝐬𝐭𝐫𝐮𝐜𝐭𝐢𝐨𝐧)

𝐀𝐫𝐭𝐢𝐜𝐥𝐞 𝐖𝐫𝐢𝐭𝐭𝐞𝐧 𝐁𝐲 - 𝐌𝐚𝐧𝐚𝐫 𝐁𝐨𝐮𝐚𝐢𝐬𝐬

𝐂𝐨𝐮𝐧𝐭𝐫𝐲 - 𝐀𝐥𝐠𝐞𝐫𝐢𝐚

Our brains are one of the most complex systems that ever existed. It is impressive how all our thoughts, memories, feelings, and even pain are all located in one place. Scientists and engineers have been studying the brain for decades yet their understanding of Bain is still limited with many mysteries that remain unsolved. However, the recent success of deep neural networks provides a research field “Brain Decoding” one of the most popular topics in neuroscience nowadays.

Interesting research in that field called “Deep image reconstruction from human brain activity”, which has been used to reconstruct images from the brain with convolution neural networks to study how the brain processes static images and other visual stimuli.

One goal of decoding neural activity is to understand brain organization by identifying which parts of the brain encode which type of information, more concretely, decoding neural activity is a method to better understand specific regions of the brain, using functional magnetic resonance imaging (fMRI) which measures brain activity by identifying areas that are being fed with oxygenated blood, which light up as colored blobs in the scans, this identifies the area in the brain responsible of processing the corresponding activity.

A person looking at an image while his brain activity as measured by fMRI, data is constructed from this measurement, and then it is used as an input to the generator, this generator is trained to reconstruct the images to be as similar as possible to the training (original) images.

The image seen by the subject can be reconstructed by using visual cortex activity, and then both the reconstructed and true training images are used as input to the comparator, which compares the image similarity in feature space, this helps optimize the reconstruction to be more accurate and realistic. These recordings are fed into a 'pattern classifier', a computer algorithm that learns the patterns associated with each picture or concept, once the program has seen enough samples, it can start to deduce what the person is looking at and generate images following the patterns that it had been trained on before, however, the accuracy of this model is still not very high and requires a lot more data to be fed into the model to enhance its performance.

Brain decoding techniques are developing so fast, although these researchers are still limited and challenging, it is certainly a significant step towards solving brain mystery, and it leads to different insights into brain function, which opens a new era for humans to understand their brains with the help of technology.