Abstract

HANDWRITTEN MEDICAL TEXT RECOGNITION USING REGIONBASED CRNN AND CONNECTIONIST TEMPORAL CLASSIFICATION

Sandeep Yadav*, Samender Singh and Pushkal Kumar Shukla

.

Abstract

Handwriting is the process of presenting an idea or information by writing. However, as the passage of time has progressed, it has become increasingly evident that the ratio of practicing physicians to the population they serve has diminished significantly, thereby leading to a widespread notoriety among medical practitioners for their notoriously poor and often illegible cursive handwriting, a phenomenon that has, rather surprisingly, been met with a considerable degree of societal acceptance and tolerance.The legibility problem of medical documents written in hand, especially those by physicians, has been one of the biggest issues in health care for decades. This paper presents an innovative solution to the issue of illegible doctor's handwriting in medical records, Doctor's Handwriting Recognition. Our idea has exceeded its function as a recognition system, which is proof that technology can bring together tradition and innovation in healthcare documentation. In the first place, it is necessary to digitize medical records for the sake of better patient care, optimal operations, and protection of data. The recognition system utilizes a Region-based deep Convolutional Neural Network (R-CRNN) that is augmented by the Connectionist Temporal Categorical (CTC) loss function. This configuration enables the system to accommodate the unique handwriting styles of individual physicians. The Doctor's Handwriting Recognition technology has the potential to transform how healthcare professionals engage with handwritten medical data. It promotes greater efficiency, bolsters patient safety, and reduces the incidence of medical errors. Consequently, this technological innovation contributes to better healthcare documentation and improves the accessibility of medical records, ultimately benefiting.

Keywords: Doctor’sHandwriting Recognition Connectionist Temporal Categorical (CTC) lossDeep Learning (DL), Image Segmentation, (R-CRNN)patient well-being.