Challenges in the Automated Diagnosis of Parkinson’s Disease across various modalities

Parkinson’s disease (PsD) is a neuro-degenerative disorder whose actual cause is unknown. PsD is the second most common brain disease worldwide, yet its diagnosis is difficult, involving multiple visits to clinicians and continuous monitoring and tests that require a lot of resources and time. So, to make the diagnosis and treatment accessible and affordable for the masses, a lot of research is being done to automate the diagnosis of the disease. Due to the slow, progressive, and irreversible nature of the disease and its symptoms overlapping with other diseases, early automated diagnosis is an active research area. Early detection can slow the progression of PsD and symptoms can be managed well, making daily life comfortable for patients. This paper aims to identify the deficits in these automated PsD detection tools which prohibit them from being incorporated into the clinical settings. Various research gaps are identified in the automated diagnosis of PsD using 8 different modalities like Magnetic Resonance Imaging (MRI), Single-photon emission computed tomography (SPECT), Positron emission tomography (PET), handwriting, gait, speech, face recognition, and Electroencephalogram (EEG). It also suggests the applications of our findings and future scope in the field of PsD diagnosis.