Israel Medical Association Journal

Ovarian cancer is a major cause of cancer death among women worldwide, and particularly in Israel. Although the disease at stage IA has 5 year survival rates of over 90%, early detection methods are not sufficiently accurate. Consequently, ovarian cancer is typically diagnosed late, which results in high fatality rates. An excellent candidate for early ovarian cancer detection would be in vivo magnetic resonance spectroscopy (MRS) because it is non-invasive and free of ionizing radiation. In addition, it potentially identifies metabolic features of cancer. Detecting these metabolic features depends on adequate processing of encoded MRS time signals for reconstructing interpretable information. The conventional Fourier-based method currently used in all clinical scanners is inadequate for this task. Thus, cancerous and benign ovarian lesions are not well distinguished. Advanced signal processing, such as the fast Padé transform (FPT) with high-resolution and clinically reliable quantification, is needed. The effectiveness of the FPT was demonstrated in proof-of-concept studies on noise-controlled MRS data associated with benign and cancerous ovaries. The FPT has now been successfully applied to MRS time signals encoded in vivo from a borderline serous cystic ovarian tumor. Noise was effectively separated out to identify and quantify genuine spectral constituents that are densely packed and often overlapping. Among these spectral constituents are recognized and possible cancer biomarkers including phosphocholine, choline, isoleucine, valine, lactate, threonine, alanine, and myoinositol. Most of these resonances remain undetected with Fourier-based in vivo MRS of the ovary. With Padé optimization, in vivo MRS could become a key method for assessing ovarian lesions, more effectively detecting ovarian cancer early, thereby improving survival for women afflicted with this malignancy.

There are major dilemmas regarding the optimal modalities for breast cancer screening. This is of particular relevance to Israel because of its high-risk population. It was suggested that an avenue for further research would be to incorporate advances in signal processing through the fast Padé transform (FPT) to magnetic resonance spectroscopy (MRS). We have now applied the FPT[1] to time signals that were generated according to in vitro MRS[2] data as encoded from extracted breast specimens from normal, non-infiltrated breast tissue, fibroadenoma and cancerous breast tissue. The FPT is shown to resolve and precisely quantify the physical resonances as encountered in normal versus benign versus malignant breast. The FPT unambiguously delineated and quantified diagnostically important metabolites such as lactate, as well as phosphocholine, which very closely overlaps with glycerophosphocholine and phosphoethanolamine, and may represent a magnetic resonance-visible molecular marker of breast cancer. These advantages of the FPT could clearly be of benefit for breast cancer diagnostics via MRS. This line of investigation should continue with encoded data from benign and malignant breast tissue, in vitro and in vivo. We anticipate that Padé-optimized MRS will reduce the false positive rates of MR-based modalities and further improve their sensitivity. Once this is achieved, and given that MR entails no exposure to ionizing radiation, new possibilities for screening and early detection emerge, especially for risk groups. For example, Padé-optimized MRS together with MR imaging could be used with greater surveillance frequency among younger women with high risk of breast cancer.

Israel has a National Screening Program for early detection of breast cancer. The need to continue and even expand this program was recently stressed in light of the high risk in the population. However, the optimal modalities for breast cancer screening are controversial, especially for women at risk. Mammography, the established screening method, is critically examined, and molecular imaging techniques, such as magnetic resonance spectroscopy and spectroscopic imaging are explored, especially for primary breast cancer detection. MRS[1] and MRSI[2] are currently limited by their reliance on the conventional framework for data analysis in biomedical imaging, i.e., the fast Fourier transform. Recent mathematical advances in signal processing via the fast Pade transform can extract diagnostically important information, which until now has been unavailable with in vivo MRS. A clinical MRS signal illustrates the rapid and stable convergence provided by FPT[3], yielding accurate information about key metabolites and their concentrations at short acquisition times. We suggest that the next step would be to apply the FPT to in vivo MRS/MRSI signals from patients with breast cancer and to compare these to findings for normal breast tissue. The potential implications of such an optimized MRS/MRSI for breast cancer screening strategies are discussed, especially for younger women at high risk.