International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 10 Issue: 05 | May 2023
p-ISSN: 2395-0072
www.irjet.net
Medical Prescription Scanner using Optical Character Recognition Swapnil Ghosh1, Vedang Pokharkar2, Sanjeevani Parida3, Saiem Pathan4, Prof. Palash Sontakke5 1,2,3,4 Undergraduate Students, Dept. of Information Technology, MIT SOC, MIT Art Design Technology University,
Pune, India
5Professor, Dept. of Information Technology, MIT SOC, MIT Art Design Technology University, Pune, India
---------------------------------------------------------------------***--------------------------------------------------------------------2. MOTIVATION Abstract - We aim to assist people in the process of prescription management and referencing. For any person, the process of going to the hospital involves receiving a single or multiple prescriptions, said prescriptions aren’t always standardized and vary from hospital to hospital, some might be handwritten and others might be digital. This makes the process of handling these documents rather tedious and complicated as most people either lose track or are forced to maintain a file for these documents. This is the problem that we aim to solve, where using deep learning, OCR (Optical Character Recognition) and handwriting recognition, we can digitize and store prescriptions thus liberating the user from their hardship by introducing a medical prescription manager that stores all the prescriptions. All the user has to do is scan the document once using the camera of their device. How does one achieve this? By the implementation of deep learning techniques and algorithms such as ’xyz’, we can effectively identify characters and whitespaces whether handwritten or digital and using a predefined vocabulary of medicinal terms accurately predict the terms such as name of medicine, dosage, etc. thus proving itself useful for the patient as well as doctors and pharmacists. Key Words: scanning
Since time immemorial, the relationship between doctor and patient has included a prescription as an intermediary for the benefit of the patient. It is through this prescription that doctors communicate the remedy of the patient’s ailment to not just the patient but also the pharmacist who is going to provide those medicines and to other doctors that the patient may ever visit. The problem arises when someone who needs to visit the doctor frequently has to keep track of their medicines and allergies. The higher the number of visits, the greater the volume of the paper trail that needs to be maintained. It slowly becomes a very cumbersome task to store all these documents as they become the backbone of the medical history and record of the patient. Even in the modernized 21st century most establishments still rely on traditional methods to generate these prescriptions. To provide a solution, we can use the help of OCR.
3. METHODOLOGY 3.1 GATHERING DATA The first step is to collect/use a large dataset of medical prescriptions that includes a variety of handwriting styles, a dictionary of medical terminology, and prescription formats. This dataset will be used to train and test the OCR model.
OCR, handwriting, medicine, prescription,
1. INTRODUCTION
3.2 PRE-PROCESSING
Every visit to the doctor’s office involves the generation of one or multiple prescriptions. For those who frequent the doctor’s office often, the prospect of keeping track of yet another prescription is dreadful as it needs to be maintained for a long duration to avoid any side effects in case new medicines that have been prescribed generate an unforeseen reaction. From the point of view of the doctor or the hospital at large, they also need to maintain comprehensive records of the hundreds if not thousands of patients that visit them, this makes it exceptionally tedious to maintain all these records. Prescriptions in general are a rather archaic solution, while some hospitals have transitioned into using digital prescriptions, they still need to print them out while solving the problem only in a limited capacity, that of the indistinguishable handwriting of the doctors while still contributing to the paper trail. As OCR is capable of identifying both handwritten and digital writings, it solves both problems at once.
© 2023, IRJET
|
Impact Factor value: 8.226
To ensure accurate recognition of medical prescriptions using OCR technology, the collected data must undergo preprocessing to eliminate unwanted elements such as background noise, stains, and irrelevant marks. The preprocessing step includes various techniques to ensure that the prescription is aligned correctly and ready for accurate OCR recognition. This step is crucial in enhancing the quality of the image and reducing noise, leading to improved OCR performance. Some of the basic techniques used are: 1) Normalization: It is a pre-processing technique in Handwritten OCR that aims to standardize the size, orientation, and position of the input image. This technique helps to make the input images more consistent and uniform, which can lead to improved OCR accuracy. The process of
|
ISO 9001:2008 Certified Journal
|
Page 1208