Multimodal Data Preprocessing Techniques for Automated Ectopic Pregnancy Risk Analysis Using Deep Learning

DOI

10.2991/978-94-6239-616-6_62How to use a DOI?

Keywords

Ectopic Pregnancy; Multimodal Data; Ultrasound Imaging; Data Preprocessing; Median Filtering; Normalization; β-hCG Interpolation; Deep Learning; Peak Signal-to-Noise Ratio

Abstract

Ectopic pregnancy is a life-threatening condition that requires urgent and accurate diagnosis or management to avoid dire consequences. The success of any predictor or diagnosis model relies on the quality and consistency of input data. This research describes the preprocessing step of a multimodal deep learning framework for ectopic pregnancy detection, recurrence prediction, and treatment suggestions. The Preprocessing of multimodal data comprises of ultrasound images, clinical data, and hormonal β-hCG level data, where each data modality is processed using its own preprocessing pipeline. For ultrasound images, all images were converted to grayscale, denoised using a median filtering algorithm to improve visual clarity and enhance structural characteristics of the gestational sac. Preclinical and clinical histories data underwent missing value imputation, outliers removal, and Min-Max normalization to assure that features were consistent and data were not heavily skewed. The hormonal data linear interpolation was used for β-hCG series to fill in the missing time points. The applied preprocessing methods have led to a significant improvement of the ultrasound image quality, where PSNR was raised from 31.18 dB to 38.18 dB, and MSE was lowered from 49.87 to 15.00. Also, the preprocessing of clinical data resulted in stable feature ranges with a uniform distribution, and the hormonal data were also preprocessed by temporal smoothing and z-score normalization for noise removal and inter-patient variability, to make the data smooth with clinically meaningful trends over time. The preprocessing is a significant step for establishing a robust pipeline for feature extraction and classification, to ensure that downstream model can produce optimal diagnosis which is also a crucial step in increasing the accuracy of the system.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - M. Maragadhavalli Meenakshi
AU  - J. Persis Jessintha
PY  - 2026
DA  - 2026/03/31
TI  - Multimodal Data Preprocessing Techniques for Automated Ectopic Pregnancy Risk Analysis Using Deep Learning
BT  - Proceedings of the International Conference on Artificial Intelligence and Secure Data Analytics (ICAISDA 2025)
PB  - Atlantis Press
SP  - 826
EP  - 836
SN  - 1951-6851
UR  - https://doi.org/10.2991/978-94-6239-616-6_62
DO  - 10.2991/978-94-6239-616-6_62
ID  - Meenakshi2026
ER  -