Multiscale wavelet filter for high resolution radioisotopic images

Radioisotopic Investigations are very useful for functional diagnosis, but available gamma cameras have poor spatial resolution. Development of new PSPMT based high-resolution detectors brings new interests to this medical imaging technique. However, the Poissonian nature of emitted photon flux and low count images due to low doses administered to the patients and to collimation, lead to images with granular aspect. This noisy appearance is particularly inconvenient when looking for cold nodules as in thyroid examinations. Low-pass filters, which reduce noise, are not well suited for detecting small cold nodules. An adapted translation invariant wavelet filter is designed which takes into account the poissonian feature of noise and preserves small sized structures. The filter was tested by Monte Carlo simulations, phantom and clinical thyroid scintigraphies. Cold spherical nodules as small as 4 mm in diameter are easily detected on filtered images acquired with IRIS (ARIES) high-resolution gamma ray imaging probe. Presentation of filtered images at different scales offering enhanced visual quality and lowered noise, simultaneously with original image, allows more comfortable analysis and improves scintigraphic diagnosis ability.