Spatial–Temporal Cube Denoising for Real-Time Digital PCR Melting Analysis to Improve the Accuracy of Multiplex Detection

Abstract

Real-time digital melting curves combine highly sensitive real-time digital polymerase chain reaction (PCR) with high-resolution melting curve analysis to achieve multiplex detection, which optimizes PCR efficiency and improves identification capability. However, due to the noise interference during the experiment, it is challenging to accurately obtain the melting temperature by extracting the microwell signal only from a single image at each temperature, further affecting the accuracy and resolution of multiplex detection. In this work, a spatial–temporal cube denoising model (STCDM) was established, which explicitly integrates the spatial and temporal dimensions to address the noise inherent in melting images. By constructing a three-dimensional spatial–temporal cube, the STCDM performs block denoising to effectively mitigate noise across both dimensions, leading to more accurate and reliable multiplex detection. The correction results demonstrated an improvement in melting temperature accuracy from 92% to 98%, with a resolution within 0.6 °C and good repeatability. Therefore, on the basis of the real-time dPCR platform, using the STCDM can significantly enhance the accuracy, driving the advancement of multiplex detection technology.