Chapter I: Introduction
In the United States, Prostate cancer (PCa) is the most common non-cutaneous malignancy after lung cancer in men with an incident of 126 per 100,000 people. The American Cancer Society (ACS) reported that the number of new cases of men diagnosed with PCa by 2016 will be an approximately 180,890 cases. PCa is still considered as the second leading cause of cancer death in men. Also, the ACS added that the estimation for death will occur from PCa is 26,120 deaths in 2016. Today, many studies stated that mortality rate is gradually decreasing compared with the past years, and this due to the developing of screening the prostate cancer using serum prostatic specific antigen (PSA) and transrectal ultrasound (TRUS) in which they allow the detection of the prostate cancer earlier and stage identification.
The screening for PCa has been improved when trans-rectal ultrasound (TRUS) guided prostate biopsy is employed which provide histopathological examination, and the positive results indicate a clinical suspicion of PCa (Li, et al., 2013). But they also provide limited information on the extent and the differentiation of PCa. It is expected that the number of patients who are diagnosed with prostate cancer will double in the future. Consequently, there is a need for an effective technique to aid the physicians first for an accurate diagnose and the patients for best treatment decision.
Among all the imaging modality the availability of Magnetic Resonance Imaging (MRI) provides a solution as an excellent imaging modality to detect, localize, and stage the aggressiveness of PCa (Bloch, Lenkinski, & Rofsky, 2008). However, MRI still need for an external hardware to strengthen the utility of this imaging modality Furthermore, the application of texture analysis (TA) is a promising tool to assess PCa which has an advantage in early detection and measurement of the disease without the increase of the acquisition time.
Chapter II: Background
In 1853, John Adams, a surgeon at The London Hospital discovered the first prostatic cancer case as a rare disease in a 59-year-old male patient by histological examination (Li, et al., 2013). 150 years later, the number of cases diagnosed with prostate cancer has increased remarkably and became a significant health problem in the US with 180,000 new cases and about 31,000 deaths occurring annually (Samuel R. Denmeade, 2002). PCa is one of the cancers that can slowly grow that they can be not threatening to patient’s life. Conversely, if its metastasize, the disease will become fatal and threaten the lives as there is currently no cure (Bloch, Lenkinski, & Rofsky, 2008).
PCa staging relies on three factors tumor, node, and metastasis (TNM) staging (Li, et al., 2013). PCa begins within the prostate gland, and it is most successfully treated when the tumor has still enclosed the gland. As mentioned earlier, once the tumor extends outside the prostate, the chances of cure diminished significantly. There are different cell lines, some of which multiply more quickly than others and tend to mutate into a more aggressive disease (Feng, 2015). Moreover, these cells have a higher probability of reaching the firm outer edge of the gland, called the capsule, and breaking through it (Feng, 2015). This capsule is called extracapsular extension (ECE), so the tumor has penetrated the capsule and begun growing outside of it (Feng, 2015). ECE is not often determined by ultrasound imaging (USI), or by TRUS biopsy (Feng, 2015). Rather, the probability of ECE was predicted by using nomograms based on clinical factors such as age, PSA and Gleason score drawn from large-scale population data (Feng, 2015). Without known evidence of ECE, many patients with both favorable and unfavorable clinical factors chose to be treated by radical prostatectomy, usually based on their doctors’ recommendations (Feng, 2015). Once the gland is removed and sent to pathology for examination, the discovery of ECE is not a pleasant revelation (Feng, 2015).
Compared with organ-confined disease, prostate cancer with extracapsular extension is associated with decreased overall and cancer-specific survival following radical prostatectomy (Feng, 2015). Patients found to have ECE are typically be sent for a course of beam radiation and/or be put on androgen deprivation therapy as a management strategy though it is not curative (Feng, 2015). Thus, the detection and treatment plan are still in a dilemma as prostate cancers demonstrate a broad range of biologic activity with the majority of a case not leading to a prostate specific death.
Also, the current treatment options available for patients have significant side effects such as incontinence, rectal injury, and impotence. Patient’s survival depends on the type of tumor is determined. In 2000, MRI in detecting ECE began to be noticed in many medical journals. MRI is highly sensitive and becomes a determining factor to provide pretreatment information and helping the physician in the treatment decision. Furthermore, recent advances in MRI/MRS of the prostate are beginning to meet these challenges.
Clinical staging without imaging:
Currently, elevated levels of the PSA in the blood and DRE are still used for diagnosing and staging prostatic cancer. PSA has been approved by the U.S Food and Drug Administration (FDA) in 1986 to be used as a primary biomarker to diagnose PCa (Li, et al., 2013). It has some limitation in which it is not capable of differentiating between PCa and benign prostatic hyperplasia (BPH). Thus, recent advances in medical imaging such as MRI and others are beginning to meet these challenges.
Imaging of prostate cancer:
Ultrasound Imaging (USI):
High PSA levels typically indicate for a blinded sextant TRUS-guided symmetrical needle biopsy. However, TRUS biopsies have been associated with a significantly lower Cap detection accuracy due to the low specificity of the PSA and poor image resolution of ultrasound. USI has limited value to its limitation to its spatial resolution and not proven satisfactory for local staging of prostate cancer (American College of Radiology, 2012). Furthermore, prostate cancer appears as a hypoechoic lesion in the peripheral zone on the transrectal probe. Many cancers can be undetected and are presumably isoechoic. A study of 2427 men provided through the American Society, a total of 52 cancers were detected. Of these, TRUS identified 44 (85%), indicating its limited sensitivity. The addition of color and power Doppler has been reported to improve the detection of prostate cancer by identifying increased vascularity but has not yet been shown to improve staging accuracy (American College of Radiology, 2012).
Computed Tomography (CT)
CT lacks sufficient soft tissue contrast in initial staging and in assessing the local extent of prostatic carcinoma in low-to-intermediate –risk patients. However, it has a great value in the evaluation of distant spread of the disease and should be reserved for use in patients with higher probability of metastases (Heath, 1998).
Magnetic Resonance Imaging MRI
Magnetic Resonance Imaging has been employed in the development of the noninvasive approach to assess and detects prostate cancer because it provides the highest spatial resolution compared with the other imaging modalities. It can be performed with or without the insertion of the endorectal coil. Although the endorectal coil insertion discomfort some patients, using it whether in 1.5T or higher field strength has the benefit of providing the highest spatial resolution among all the imaging modalities. Furthermore, the variation of techniques including T1-weighted images, T2 weighted images, dynamic contrast enhanced (DCE) T1-weighted images (DCE-T1-WI), MR spectroscopy (MRS), diffusion weighted images (DWI), provide many chanced to diagnose the biologic processes.
T2 weighted images (T2-WI):
T2 weighted image is the most commonly used since it led to an excellent image quality and resolution. This is evidenced in the images below.
Figure 1 showing the coronal T2-Weighted Imaging of normal prostate anatomy on MRI
Source: (Abdellaoui, Iyengar & Freeman, 2011).
Figure 2 showing the Axial T2-weighted imaging of the prostate cancer for the left peripheral zone with a bugle of the capsule indicating early extracapsular invasion (arrow)
Source: (Abdellaoui, Iyengar & Freeman, 2011).
The images above show an improvement in the sensitivity of the MRI. Consequently, it is able to detect the extra-capsular extension. Due to the improved sensitivity, the MRI is also able to detect the invasion of the seminal vesicle (Abdellaoui, Iyengar & Freeman, 2011). This is because the specificity improves from the initial range of between 73% and 80% to a range of between 97% and 100%. Consequently, the extracapsular extension that has a distance of 0.5 millimeters is detected accurately (Abdellaoui, Iyengar & Freeman, 2011).
Figure 3 showing the coronal T2-Weighted imaging of prostate cancer of the prostate base with invasion of the seminal vesicles (arrow)
Source: (Abdellaoui, Iyengar & Freeman, 2011).
There are intrinsic complexities in imaging prostate that limit staging accuracy. The good spatial resolution achieved with the endorectal coil in 3T MRI scan with at least halving the voxel size (voxel size 0.35 mm3 v 0.66-1.12 mm3), reveals pathoanatomic details on T2-WI not seen at 1.5T or 3T without endorectal coil (Bloch, Lenkinski, & Rofsky, 2008). In T2-WI, the interpretation of PCa can be affected by false-positive findings such as prostatitis, post-biopsy hemorrhage, and fibrosis. Thus, the addition of functional magnetic resonance imaging (fMRI) is necessary to improve the accuracy of diagnosing PCa such as dynamic contrast enhanced-MRI (DCE-MRI), MR spectroscopy (MRS), diffusion weighted imaging (DWI) (Li, et al., 2013).
Dynamic Contrast Enhanced MRI (DCE- MRI):
DCE-MRI can diagnose earlier and more intense enhancement in sites of tumor compared with the normal peripheral zone. Prostate cancer like any tumors has two factors microvessel density (MVD) and angiogenic that were evaluated as prognostic factors in patients with prostate cancer. MVD has been correlated with clinical and pathological stage, metastasis, and histological grade in prostate cancer (Bloch, Lenkinski, & Rofsky, 2008). Although there is some controversy, MVD also has been correlated with disease-specific survival and progression after treatment. Moreover, recent data suggest that DCE-MRI can provide valuable information about individual MVD in prostate cancer. Thus, there are biological features associated with prostate cancer that can be demonstrated with DCE-MRI for further disease characterization (Bloch, Lenkinski, & Rofsky, 2008).
Magnetic Resonance Spectroscopic Imaging (MRSI):
The addition of MRSI to MR imaging significantly improves characterization of peripheral zone prostate tissue as benign or malignant. Coakley (2003) demonstrated that prostate cancers have a characteristic loss of the citrate peak and gain in the choline/creatine peak on MRSI. Moreover, the ratio of choline to citrate is related to the Gleason score, indicating that MRSI provides information about tumor aggressiveness (American College of Radiology, 2012). MRS facilitates the differentiation of normal and altered tissue metabolism. Therefore, it is different to other imaging methods that only assess abnormalities of structure or perfusion. Incremental improvement in accuracy of cancer detection and staging has been reported when MRSI was added to endorectal MRI (erMRI) alone (American College of Radiology, 2012).). As an indicator of outcome, MRSI has been shown predictive of biochemical recurrence. However, a recent American College of Radiology Imaging Network® (ACRIN®) multicenter trial showed no incremental benefit of MRSI for localizing prostate cancer over 1.5T erMRI alone (American College of Radiology, 2012). MRSI cannot yet be considered to provide significant advantages in local staging before treatment (American College of Radiology, 2012).
Diffusion-weighted imaging (DWI):
The inclusion of DWI technique to MR prostate imaging gives an additional method to improve prostate tumor detection and localization compared to T2-Weighted Images alone (American College of Radiology, 2012). It generates tissue contrast reflecting water molecular diffusion using apparent diffusion coefficient (ADC) mapping. MR diffusion has been used commonly for evaluating acute stroke in the brain. Recently, it has been suggested that DWI may also play a role in the early detection of tumor response to therapy.
Although MRI in managing and detecting the prostate cancer is expanding, its exact role is not yet defined. The staging of PCa is still challenging in which there are potential indications for prostate MRI include surveillance in patients known to have low-risk prostate cancer, for staging in patients with intermediate and high-risk cancer prior to therapy, and for detection of cancer in patients with elevated PSA but negative TRUS biopsy (Sajal & Pokharel, 2015).
Benign prostatic hyperplasia (BPH) is not a prostate cancer. It is a benign enlargement of the prostate due to an abnormal growth of the noncancerous prostate cells. They differ in the way they develop. Prostate cancer commences in the outer peripheral zone of the prostate and grown outward invading the surrounding tissues whereas in BPH the growth in inward toward the prostate’s core and begins in the inner area of the prostate called the transition zone that is a ring tissue circling and tightening the urethra. That’s why BHP produces noticeable symptoms such as affecting the urination while the prostate cancer is often silent disease with no obvious symptoms often for years.
CAD analysis of DCE MRI ( Breast MRI prostate MRI)
The most prevalently used approach towards the screening and detection of breast cancer is the conventional X-ray mammography (Yang, Li, Zhang, Shao, Zhang & Zheng, 2014). The continued use of this approach is not due to the fact that it is very effective in different conditions. In fact, the conventional X-ray mammography is faced with several limitations. For instance, the detection sensitivity for breast cancer using this approach for screening mammography reduces significantly from 98 to 100% when the patient has fatty breasts to 30 to 48% when the patient has dense breasts (Yang et al., 2014).
The Dynamic Contrast Material-Enhanced Magnetic Resonance Imaging (DCE-MRI) is used in the screening and detection of breast cancer because it has a higher sensitivity when compared to the conventional X-ray mammography. The heightened sensitivity is advantageous in the detection of breast cancer in its early stages as well as in women who have dense breast (Yang et al., 2014). The sensitivity is increased from between 33% and 59% as is experienced when an X-ray mammography is used to 71% and 94%.
Despite the momentous benefits that the use of the Dynamic Contrast Material-Enhanced Magnetic Resonance Imaging (DCE-MRI) has brought in the screening and detection of breast cancer, there are challenges with which this approach comes. These challenges have an effect on the outcome of the screening procedures. One of these challenges as found by Yang et al., (2014) is the reduced and at times comparable specificity when the use of the Dynamic Contrast Material-Enhanced Magnetic Resonance Imaging (DCE-MRI) is pitted against the conventional X-ray mammography. Reduced specificity means that there are more false positives when the Dynamic Contrast Material-Enhanced Magnetic Resonance Imaging (DCE-MRI) is used for screening breast cancer compared to when conventional X-ray mammography is used.
The reduced specificity has a significant influence on patient outcomes. For instance, a false positive can increase the level of anxiety in a patient. Additionally, the reduced specificity causes an increase in the cost of healthcare. This is because the number of unnecessary recalls increases due to the need for more imaging and biopsy workups (Yang et al., 2014). Considering that the Dynamic Contrast Material-Enhanced Magnetic Resonance Imaging (DCE-MRI) is more sensitive that other approaches, any measures to improve its specificity would help improve the overall effectiveness of the approach in the screening and detection of breast cancer. In this regard, the Computer Aided Diagnosis has been explored as a way of improving the accuracy of the method. Yang et al., (2014) found that the use of the Computer Aided Diagnosis schemes helped improve the accuracy of the Dynamic Contrast Material-Enhanced Magnetic Resonance Imaging (DCE-MRI).
Summary, Challenges (BPH and Cancer overlap.)
The prevalence of prostate cancer necessitates a diagnostic procedure that is both sensitive and specific. There are different diagnostic procedures at the disposal of physicians. Different procedures bring with them different benefits and limitations. May of the treatment options for prostate cancer result in rectal injuries for the patient, incontinence, and impotence. This poses a challenge for the detection and treatment of prostate cancer. The use of the magnetic resonance imaging has proved valuable in the detection, localization, and the determination of the stage aggressiveness of prostate cancer. This is because it is more sensitive that the other procedures available. It helps address the challenges highlighted above in addition to providing better results when compared to the other approaches. However, it has a limitation in that it is less specific compared to conventional methods. Continued research and improvements have offered solutions which show more promise in terms of the improvement of the effectiveness of the magnetic resonance imaging in the screening, detection, and localization of prostate and breast cancer.
Texture analysis
The diagnosis of prostate cancer using the MRI is very challenging, especially during the separation of benign confounders such as inflammation, atrophy, prostatic intra-epithelial neoplasia, and benign prostatic hyperplasia and the prostate cancer. Although there is an increase interest in the role of MRI to evaluate PCa aggressiveness, there is still need for advance methods of image processing and analysis as a next step to overcome the challenges. Texture analyses is a branch of image processing which aims to reduce image information by extracting texture description from image (Barry, 2014). This extraction may allow for the mathematical and statistical analyses to detect a subtle MRI signal changes among image pixels (Barry, 2014). In other words, it determines the relationship between adjacent pixels within an image that is not seen and distinguish by human eye (Wibmer A1, 2015).
It describes how often one grey tone will appear in a specified spatial relationship to another grey tone on the image (Wibmer, 2015) (Haralick, 1973). By using a series of mathematical equations, it generates a range of quantitative parameters ‘texture features’ that characterize the spatial variation of grey levels throughout an image (Wibmer A1, 2015). The early changes in the image texture are of particular relevance, as relatively normal-appearing tissues with subtle microscopic disturbances due to disease, such as in the case of hepatic fibrosis, may be detected in its earlier stages (HeiShun, 2015). The major advantage of this approach is the ability to detect early and quantify a chronic disease without increasing the acquisition time of the image or the dose for the patient.
Texture analysis is commonly done using the MATLAB analysis software. In the first-order grey scale analysis, individual pixels are analyzed for their values. The mean values, and the difference between the minimal and maximal pixel values is determined. In the second order grey scale analysis the co-occurrence matrices is used to analyze the textural features. For instance entropy matrices determine the extent of the prostate cancer and the distribution of its intensities. Contrast matrices are used in determining the weighted mean differences of the neighboring pixels’ intensity. The correlation is used to determine the relationship between the intensity of neighboring pixels. Using these analyses, analysts can tell benign cells from cancerous cells (Nguyen et al., 2012).
References
Abdellaoui, A., Iyengar, S. and Freeman, S. (2011). Imaging in Prostate Cancer. Future Oncology. 7(5): 679-691.
American College of Radiology. (2012). ACR Appropriateness Criteria. Retrieved from https://acsearch.acr.org/docs/69371/Narrative/
Barry, K. (2014). Quantifying Liver Fibrosis Through The Application of Texture Analysis to Diffusion Weighted Imaging. Magnetic Resonance Imaging , 84-90.
Bloch, N. B., Lenkinski, R., & Rofsky, N. (2008). The role of magnetic resonance imaging (MRI) in prostate cancer imaging and staging at 1.5 and 3 Tesla: the Beth Israel Deaconess Medical Center (BIDMC) approach. Cancer Biomark , 251-262.
Coakley FV, Q. A. (2003). Magnetic Resonance Imaging and Spectroscopic Imaging of Prostate Cancer. Journal of Urology , 69-75.
Feng T. (2015). Multiparametric Magnetic Resonance Imaging Localizes Established Extracapsular Extension of Prostate Cancer. Urologic Oncology, 33 (3), 109.e15– 109.e22.
Haralick R. (1973). Textural Features for Image Classification. IEEE Trans Syst Man Cybern , 610–621.
Heath, M. M. (1998, May 15). Chronic Bronchitis: Primary Care Management. American Family Physician , 2365-2372.
HeiShun Yu, K. B. (2015). Utility of Texture Analysis for Quantifying Hepatic Fibrosis on Proton Density MRI. Journal of Magnetic Resonance Imaging , 1259–1265.
Li, L., Wing, L., Feng, Z., Hu, Z., Wang, G., Yuan, X., et al. (2013). Prostate Cancer Magnetic Resonance Imaging (MRI): Multidisciplinary Standpoint. Quantitative Imaging in Medicine and Surgery , 100-112.
Nguyen, P., Hundloe, J., Salvado, O., Dowson, N., Ware, R., Masters, I., Bhatt, M., Kumar, A. and Fielding, D. (2012). Optical Differentiation Between Malignant and Benign Lymphadenopathy by Grey Scale Texture Analysis of Endobronchial Ultrasound Convex Probe Images. Chest. 141(3): 709-715.
Sajal S. & Pokharel, N. U. (2015). Multi-Parametric MRI Findings of Transitional Zone Prostate Cancers: Correlation with 3D Transperineal Mapping Biopsy. Abdomen Imaging , 143– 150.
Wibmer, H. H. (2015). Haralick Texture Analysis of Prostate MRI: Utility for Differentiating Non-Cancerous Prostate from Prostate Cancer and Differentiating Prostate Cancers with Different Gleason Scores. European Radiology , 2840-2850.
Yang, Q., Li, L., Zhang, J., Shao, G., Zhang, C. and Zheng, B. (2014). Computer-Aided Diagnosis of Breast DCE-MRI Images Using Bilateral Asymmetry of Contrast Enhancement between Two Breasts. Journal of Digital Imaging. 27(1): 152-160.
