Optimization of Brain Tumor MR Image Classification Accuracy Using Optimal Threshold, PCA and Training ANFIS with Different Repetitions.

BACKGROUND: One of the leading causes of death is brain tumors. Accurate tumor classification leads to appropriate decision making and providing the most efficient treatment to the patients. This study aims to optimize brain tumor MR images classification accuracy using optimal threshold, PCA and training Adaptive Neuro Fuzzy Inference System (ANFIS) with different repetitions. MATERIAL AND METHODS: The procedure used in this study consists of five steps: (1) T1, T2 weighted images collection, (2) tumor separation with different threshold levels, (3) feature extraction, (4) presence and absence of feature reduction applying principal component analysis (PCA) and (5) ANFIS classification with 0, 20 and 200 training repetitions. RESULTS: ANFIS accuracy was 40%, 80% and 97% for all features and 97%, 98.5% and 100% for the 6 selected features by PCA in 0, 20 and 200 training repetitions, respectively. CONCLUSION: The findings of the present study demonstrated that accuracy can be raised up to 100% by using an optimized threshold method, PCA and increasing training repetitions.

In vivo imaging of brain cancer using epidermal growth factor single domain antibody bioconjugated to near-infrared quantum dots.

Diagnosis of glioblastoma multiform (GBM) with MRI lacks molecular information and requires a biopsy for pathologic confirmation. The EGFRvIII, is a constitutively active mutant of the EGF receptor, identified in a high percentage of brain cancers and associated with increased invasiveness and resistance, making it a good target to improve imaging and diagnosis. The present study shows that conjugation of near-infrared quantum dot (Qd800) to an anti-EGFRvIII single domain antibody, made of the variable region with an extra cysteine for site-specific conjugation (EG2-Cys), increased its internalization in U87MG-EGFRvIII cells in vitro compared to Qd800 conjugated with the Fc region of the antibody (EG2-hFc) or unconjugated. EG2-Cys also improved the contrast in Near-Infrared Imaging of mice bearing orthotopic glioblastoma. The increased accumulation was confirmed by fluorescence microscopy of brain sections. The specificity of EG2-Cys in brain tumor expressing the EGFRvIII mutant receptor may provide an accurate less invasive diagnosis and determine the level of tumor aggressiveness and resistance.

Influence of glioma tumour microenvironment on the transport of ANG1005 via low-density lipoprotein receptor-related protein 1.

BACKGROUND: ANG1005 consists of three molecules of paclitaxel conjugated via ester bonds to the 19-amino-acid peptide Angiopep-2. The new chemical agent has been shown to cross the blood-brain barrier (BBB) by receptor-mediated transcytosis via low-density lipoprotein receptor-related protein 1 (LRP1). The experiments here examined the role of LRP1 in the subsequent endocytosis of drug into cancer cells. METHODS: Localisation of ANG1005 and Angiopep-2 was examined by immunohistochemistry and in-vivo near-infrared fluorescence imaging in mice carrying orthotopic glioma tumours. Transport of ANG1005 and Angiopep-2 was examined in U87 glioblastoma cell lines. RESULTS: Systemically administered ANG1005 and Cy5.5Angiopep-2 localised to orthotopic glioma tumours in mice. The glioma transplants correlated with high expression levels of LRP1. Decreasing LRP1 activity, by RNA silencing or LRP1 competitors, decreased uptake of ANG1005 and Angiopep-2 into U87 glioblastoma cells. Conversely, LRP1 expression and endocytosis rates for ANG1005 and Angiopep-2 increased in U87 cells under conditions that mimicked the microenvironment near aggressive tumours, that is, hypoxic and acidic conditions. CONCLUSION: ANG1005 might be a particularly effective chemotherapeutic agent for the wide array of known LRP1-expressing brain and non-brain cancers, in particular those with an aggressive phenotype.

Incidence of acute peripheral neurotoxicity after deep regional hyperthermia of the pelvis.

BACKGROUND: After observing rather severe acute neurotoxicity in a few patients following deep hyperthermia treatment for a pelvic tumour, we evaluated the incidence of neurotoxicity in all patients treated with deep hyperthermia of the pelvis between June 1990 and April 2004. MATERIALS AND METHODS: Hyperthermia treatment registrations and hospital charts of all 736 patients were reviewed. Differences between the incidence of neurotoxicity in subgroups of patients were evaluated by 2 x 2 exact tests. RESULTS: Grade 2 or 3 acute neurotoxicity occurred in 2.3% of patients, grade 3 in 0.7%. The duration of symptoms was longer than 3 months in 6 patients (0.8%). Neurological examination in 5 patients showed that the most commonly involved structures are the sacral and lower lumbar nerve roots and the sacral plexus. Acute neurotoxicity occurred only after November 1999 and only in patients treated for primary cervical cancer. Comparison of applied powers and achieved temperatures in patients developing neurotoxicity did not show differences between treatment sessions which resulted in neurotoxicity and sessions not resulting in neurotoxicity. CONCLUSION: Acute neurotoxicity following hyperthermia for pelvic tumours is a rare complication, but can result in symptoms affecting the activities of daily life. We found no patient, tumour or treatment characteristics predictive for a risk of neurotoxicity.

Intra-luminal thermometry: is tissue type assignment a necessity for thermal analysis?

INTRODUCTION: Tissue type assignment, i.e. differentiation tumour from normal tissue, is a normal procedure for interstitial thermometry. In our department, thermometry in patients with a tumour in the lower pelvis is usually restricted to the intra-luminal tracks. It is unknown whether discrimination between normal and tumour tissue is relevant for deep regional hyperthermia thermal dosimetry using only intra-luminal tumour contact and tumour adjacent thermometry. This study has analysed the acquired temperature data in order to answer this question. PATIENTS AND METHODS: Seventy-five patients with locally advanced cervical carcinoma were selected randomly. Patients were treated with a two or three modality combination, i.e. radiotherapy +hyperthermia or radiotherapy + hyperthermia + chemotherapy from October 1997 to September 2003. The first 100 hyperthermia treatments fulfilling the only selection criterion: no displacement of the thermometry catheter along the insertion length during the treatment, were included in the study, resulting in 43 patients with one-to-five treatments/patient (median 2). Using RHyThM (Rotterdam Hyperthermia Thermal Modulator), for each single treatment tissue type, was defined on the basis of information given by a CT scan in radiotherapy position. A step change in the slope of the profile of the first temperature map was identified to verify the insertion length of the catheter. RESULTS: The average T50 (median temperature) in bladder tumour indicative, vagina tumour contact and rectum tumour indicative was 40.9 +/- 0.9 degrees C, 39.7 +/- 0.9 degrees C and 40.6 +/- 0.8 degrees C, respectively. The average normal tissue T50 in bladder, vagina and rectum was 40.8 +/- 0.9 degrees C, 40.1 +/- 0.9 degrees C and 40.7 +/- 0.8 degrees C, respectively. The differences between bladder tumour indicative T50 and bladder normal tissue T50 and also between vagina tumour contact T50 and vagina normal tissue T50 were significant ( p = 0.0001). No statistical difference was found between rectum tumour indicative t50 and rectum normal tissue T50. CONCLUSION: At present the cause of the temperature difference is not known. However, as the difference between tumour (indicative/contact) and normal tissue is very small and considering also the inaccuracy in the tissue type assignment it can be stated that this study does not provide sufficient evidence to conclude that the statistical difference has clinical relevance. Therefore, it was concluded that at this time there is no need to differentiate between normal and tumour tissue in intra-luminal thermometry.

Temperature data analysis for 22 patients with advanced cervical carcinoma treated in Rotterdam using radiotherapy, hyperthermia and chemotherapy: a reference point is needed.

INTRODUCTION: The growing interest and participation in multi-institutional trials involving deep hyperthermia treatment is an important step towards the further consolidation of hyperthermia as an oncological treatment modality. However, the differences in the clinical procedures of hyperthermia application also raises questions as how to compare the reported temperatures data obtained by the different institutes. In this study our recent developed approach, RHyThM (Rotterdam Hyperthermia Thermal Modulator), has been used for thermal data analysis to investigate the temperature dynamics behaviour of a series of deep hyperthermia treatments. PATIENTS AND METHODS: All 22 patients (104 hyperthermia treatments) with locally advanced cervical carcinoma who participated in a feasibility study for treatment with a three-modality therapy were selected. The patients received mega-voltage external beam radiotherapy to the pelvis in daily fractions of 2 Gy five times a week to a total dose of 46 Gy and additional brachytherapy, at least four courses of weekly cisplatin (40 mg m-2) and five sessions of weekly loco regional deep hyperthermia treatments with the BSD2000-3D with the Sigma 60 or the Sigma-eye applicators at frequencies 70-120 MHz. Using RHyThM tissue type was defined along the insertion length, based on the CT scan information in radiotherapy position, for each single treatment. A step change in the slope of the profile of the first temperature map was identified to verify the insertion length of the thermometry catheter and precise location of the transition between in- and outside the body. Data analysis was performed based on the temperature readout provided by RHyThM. RESULTS: The temperature and RF-power data of 97 treatments could be analysed. The intra-vaginal temperature indices were slightly lower than those for bladder and rectum. The average T50 (median temperature) in all lumens, i.e. bladder, vagina and rectum, was 40.4 +/- 0.6 degrees Celsius. The average vagina all lumen T50 was 40.0 +/- 0.8 degrees Celsius. The average bladder and rectum all lumen T50 was 40.6 +/- 0.7 degrees Celsius and 40.5 +/- 0.6, respectively. When the analysis was restricted to the deepest 5 cm of the vagina lumen, the average T50 was 39.8 +/- 0.9 degrees Celsius. Good correlation exists between the various temperature indices like T20, T50 and T90, for all lumen measurements in bladder, vagina and rectum. No correlation was found between temperature indices and treatment number. For the complete patient population, no relationship was found between T50 and net integrated RF-power applied. In an explorative analysis on individual patients a positive correlation coefficient or trend was found in 14 patients between normalized net integrated RF-power and vagina T50. CONCLUSION: Average all lumen T50 for bladder, vagina and rectum differ less than 1 degrees Celsius, indicating that a large volume was heated relatively homogeneously. The vagina T50 value depends on how many measurement points are included for the analysis. In this group of patients the vagina T50 of the first treatment is not a good measure to discriminate between patients with 'heatable' and 'non-heatable' tumours. In order to compare temperature data reported by different institutes dealing with the same group of patients, one needs a strict and clear agreement on which temperature measurements or reference point(s) that should be included in the analysis.