Doctor’s Handwritten Prescription BD dataset
Description
🔗 This dataset was used in the following publication: A. R. Mia, M. A. -A. -S. Chowdhury, A. A. Mamun, A. M. Ruddra and N. T. Tanny, "A Deep Neural Network Approach with Pioneering Local Dataset to Recognize Doctor's Handwritten Prescription in Bangladesh," 2024 International Conference on Advances in Computing, Communication, Electrical, and Smart Systems (iCACCESS), Dhaka, Bangladesh, 2024. DOI: 10.1109/ACCESS.2024.10499631 💊 Prescription Word Dataset for Machine Learning This dataset was created by extracting and processing prescription images to support educational research and experimentation in machine learning models, particularly for text recognition and classification in healthcare contexts. 📸 Dataset Creation Process To transform the prescription images into a structured dataset suitable for machine learning, a specialized word detection algorithm was employed. This code segmented the prescription images into individual words, converting the data into a format that facilitates accurate recognition by ML models. Word segmentation was the first crucial step. Each word was manually screened to retain only pharmaceutical terms. All non-drug terms were discarded. Manual labelling was performed by multiple team members independently to ensure accuracy. 📊 Dataset Details Total Words Extracted: 4,680 File Format: Excel (.xlsx) and CSV (.csv) Number of Classes (Medicines): 78 📚 Drug Names Included Beklo, Maxima, Leptic, Esoral, Omastin, Esonix, Canazole, Fixal, Progut, Diflu, Montair, Flexilax, Maxpro, Vifas, Conaz, Fexofast, Fenadin, Telfast, Dinafex, Ritch, Renova, Flugal, Axodin, Sergel, Nexum, Opton, Nexcap, Fexo, Montex, Exium, Lumona, Napa, Azithrocin, Atrizin, Monas, Nidazyl, Metsina, Baclon, Rozith, Bicozin, Ace, Amodis, Alatrol, Napa Extend, Rivotril, Montene, Filmet, Aceta, Tamen, Bacmax, Disopan, Rhinil, Flamyd, Metro, Zithrin, Candinil, Lucan-R, Backtone, Bacaid, Etizin, Az, Romycin, Azyth, Cetisoft, Dancel, Tridosil, Nizoder, Ketoral, Ketocon, Ketotab, Ketozol, Denixil, Provair, Odmon, Baclofen, MKast, Trilock, Flexibac. These classes represent commonly prescribed pharmaceutical names likely to appear in handwritten prescriptions. 🧪 Usage This dataset is ideal for: Text classification Optical character recognition (OCR) Named entity recognition (NER) Deep learning model training and evaluation ⚠️ Note: This dataset is free to use for educational and research purposes only.
Files
Steps to reproduce
1. Collect handwritten prescription images from hospitals and clinics in Bangladesh with appropriate permission and anonymize any patient-identifiable information. 2. Preprocess the prescription images to improve readability by converting them to grayscale and enhancing image quality where necessary. 3. Apply a word segmentation algorithm to detect and extract individual handwritten words from each prescription image. 4. Manually review the segmented words and remove all non-pharmaceutical terms, symbols, and incorrectly segmented regions. 5. Group the remaining medicine names into their corresponding classes. 6. Perform manual annotation of each extracted word image. Multiple reviewers independently verify the labels to improve annotation quality and consistency. 7. Resize every word image to **128 × 128 pixels** while preserving the handwritten content. 8. Store the labels in both **CSV** and **Excel** formats, with each image associated with its corresponding medicine name. 9. Organize the dataset into **78 medicine classes**, each containing **60 handwritten word images**, resulting in a total of **4,680 images**. 10. For model development, split the dataset into **60% training**, **20% validation**, and **20% testing** using a stratified sampling strategy to maintain equal class distribution. 11. The resulting dataset can be used for handwritten medicine name recognition, optical character recognition (OCR), text classification, and deep learning research.
Institutions
- Ahsanullah University of Science and TechnologyDhaka Division, Dhaka