An innovative approach to classifying various phases of tuberculosis using an improved Spiking Neural Network applied to X-ray images

Radiology using chest X-rays (CXR) is a common diagnostic tool. It is the gold standard for diagnosing deadly diseases. Chest X-rays may help radiologists diagnose illness. Medication errors often occur due to the similarity in symptoms across lung illnesses. For lung difficulties in computed tomography (CXR) pictures, radiology researchers are using CAD software more. Neural networks and machine learning (ML) may save time and simplify. Our strategy for segmenting tuberculosis (TB) into various phases is clearer and more efficient than earlier studies. A comparison-spiking neural network may detect early tuberculosis (TB). SNN outperforms other training methods including previous-generation algorithms. We utilise a modernised Gabor filter to prepare pictures for processing. We used a custom Gabor filter to remove spatial anomalies from the dataset. Next, we use E-SNN for picture classification because it outperforms all other methods in reliability (98.29 + 0.47%), accuracy (98.32 + 0.73%), sensitivity (99.41 + 0.12%), specificity (98.73 + 0.68%), error rate (0.00187), and area under the receiver operating characteristic curve (99.04 We also compared different cross-validation approaches. The Grad-Cam results support the methodology. We compared the model to others using public datasets like KIT, Shenzhen, and Montgomery. The suggested technique yields better results for the Shenzhen dataset. This approach detects active, latent, normal, and primary TB better than any others.