4. Wavelet applications in medical image processing
N. Krishna Santosh, Soubhagya Sankar Barpanda · 2020
Over the past 10 years, there is a rapid evolution in modalities of biomedical imaging. They act as aid for doctors in disease identification, estimation of disease occurrence and methodological treatment, thus degree of patient care has greatly strengthened. Medical images are usually generated either by ionizing radiation such as X-rays and gamma rays or by nonionizing radiation such as ultrasound and magnetic resonance imaging (MRI) techniques. Processing of biomedical images is analogous to acquiring biomedical signal from various dimensions. It comprises improvement, analysis and display of medical images. Quick ascertainment of deadly diseases like cancer largely improves the chances for effective treatment to save patient’s life. This can be happening fruitfully only with the effective extraction of informatics from acquired medical images. Usually, after acquiring biosignals from various biomedical modalities, they have to be properly sampled and quantized for effective signal processing. To explore valuable informatics from the signal, it should have been processed through proper techniques. The selection of techniques for exploring information from medical images contributes significantly for early ascertainment of fatal diseases. Because of biomedical image’s multifaceted nature, it isʚ challenging to explore fruitful essential statistics that can be incorporated into expert systems for diagnosis. At present, though a large number of techniques are available for processing the medical images, a single technique may not be flexible to use for various fields such as quantum mechanics, geography, medicine, pattern recognition, video processing and robot vision. The invention of wavelet method has overcome major limitations in collecting informatics form image and has a lot of impact on image processing. The discrete wavelet transform (DWT) is akin to a microscope through which we can discern various components of the signal by just altering the focus. Transformation of discrete wavelet extracts and separates a signal through a collection of quadrature filters, and they have corresponding filter properties specific to corresponding mother wavelet. The content captured by the coefficients of wavelet is unique, and it is possible to rebuild the original signal with no redundancy. This tremendous nature of wavelet inspired the development of many techniques for medical signal compression based on wavelet compression theory. These techniques are vital for improving chances to explore novel diagnosis information and medical data transmission. The signals generated form medical image modality devices have high chances for corruptiondue to photon noise, device noise, digitization noise and many more noises. This type of intrinsic noises is caused by imaging modalities, which are obstacles to explore valuable data from medical images and are very difficult to remove by applying traditional filters. The DWT procedure extracts the signal with extremely minimal distortion to avoid noise. This marvelous DWT working principle can be used on various biomedical signals and are decomposed by setting a threshold to every corresponding level detailed coefficient. These extracted coefficients from digitized images which are formed by medical signals are critical for diagnosis of disease in its initial stage so that we can save the human lives.