Biomedical subjects

Martin O Leach

Publications and source records attributed to Martin O Leach.

At least 19 recordsLinked to original sources

PubMed2007

Reference tissue quantification of DCE-MRI data without a contrast agent calibration.

The quantification of dynamic contrast-enhanced (DCE) MRI data conventionally requires a conversion from signal intensity to contrast agent concentration by measuring a change in the tissue longitudinal relaxation rate, R(1). In this paper, it is shown that the use of a spoiled gradient-echo acquisition sequence (optimized so that signal intensity scales linearly with contrast agent concentration) in conjunction with a reference tissue-derived vascular input function (VIF), avoids the need for the conversion to Gd-DTPA concentration. This study evaluates how to optimize such sequences and which dynamic time-series parameters are most suitable for this type of analysis. It is shown that signal difference and relative enhancement provide useful alternatives when full contrast agent quantification cannot be achieved, but that pharmacokinetic parameters derived from both contain sources of error (such as those caused by differences between reference tissue and region of interest proton density and native T(1) values). It is shown in a rectal cancer study that these sources of uncertainty are smaller when using signal difference, compared with relative enhancement (15 +/- 4% compared with 33 +/- 4%). Both of these uncertainties are of the order of those associated with the conversion to Gd-DTPA concentration, according to literature estimates.

Biophysical Phenomena↗

PubMed2006

A test of performance of breast MRI interpretation in a multicentre screening study.

OBJECTIVES: The aim of this study was to assess the consistency and performance of radiologists interpreting breast magnetic resonance imaging (MRI) examinations. MATERIALS AND METHODS: Two test sets of eight cases comprising cancers, benign disease, technical problems and parenchymal enhancement were prepared from two manufacturers' equipment (X and Y) and reported by 15 radiologists using the recording form and scoring system of the UK MRI breast screening study [(MAgnetic Resonance Imaging in Breast Screening (MARIBS)]. Variations in assessments of morphology, kinetic scores and diagnosis were measured by assessing intraobserver and interobserver variability and agreement. The sensitivity and specificity of reporting performances was determined using receiver operating characteristic (ROC) curve analysis. RESULTS: Intraobserver variation was seen in 13 (27.7%) of 47 of the radiologists' conclusions (four technical and seven pathological differences). Substantial interobserver variation was observed in the scores recorded for morphology, pattern of enhancement, quantification of enhancement and washout pattern. The overall sensitivity of breast MRI was high [88.6%, 95% confidence interval (CI) 77.4-94.7%], combined with a specificity of 69.2% (95% CI 60.5-76.7%). The sensitivities were similar for the two test sets (P=.3), but the specificity was significantly higher for the Manufacturer X dataset (P<.001). ROC curve analysis gave an area under the curve of 0.85 (95% CI 0.79-0.92) CONCLUSIONS: Substantial variation in all elements of the scoring system and in the overall diagnostic conclusions was observed between radiologists participating in MARIBS. High overall sensitivity was achieved with moderate specificity. Manufacturer-related differences in specificities possibly occurred because the numerical thresholds set for the scoring system were not optimised for both equipment manufacturers. Scoring systems developed on one equipment software may not be transferable to other manufacturers.

Breast Neoplasms↗

PubMed2006

Minimally invasive pharmacokinetic and pharmacodynamic technologies in hypothesis-testing clinical trials of innovative therapies.

Clinical trials of new cancer drugs should ideally include measurements of parameters such as molecular target expression, pharmacokinetic (PK) behavior, and pharmacodynamic (PD) endpoints that can be linked to measures of clinical effect. Appropriate PK/PD biomarkers facilitate proof-of-concept demonstrations for target modulation; enhance the rational selection of an optimal drug dose and schedule; aid decision-making, such as whether to continue or close a drug development project; and may explain or predict clinical outcomes. In addition, measurement of PK/PD biomarkers can minimize uncertainty associated with predicting drug safety and efficacy, reduce the high levels of drug attrition during development, accelerate drug approval, and decrease the overall costs of drug development. However, there are many challenges in the development and implementation of biomarkers that probably explain their disappointingly low implementation in phase I trials. The Pharmacodynamic/Pharmacokinetic Technologies Advisory committee of Cancer Research UK has found that submissions for phase I trials of new cancer drugs in the United Kingdom often lack detailed information about PK and/or PD endpoints, which leads to suboptimal information being obtained in those trials or to delays in starting the trials while PK/PD methods are developed and validated. Minimally invasive PK/PD technologies have logistic and ethical advantages over more invasive technologies. Here we review these technologies, emphasizing magnetic resonance spectroscopy and positron emission tomography, which provide detailed functional and metabolic information. Assays that measure effects of drugs on important biologic pathways and processes are likely to be more cost-effective than those that measure specific molecular targets. Development, validation, and implementation of minimally invasive PK/PD methods are encouraged.

Antineoplastic Agents↗

PubMed2006

Prediction of clinicopathologic response of breast cancer to primary chemotherapy at contrast-enhanced MR imaging: initial clinical results.

PURPOSE: To prospectively document changes in contrast agent kinetics in patients with primary breast cancer treated with systemic chemotherapy after one or two cycles and to determine whether kinetic measures can be used to predict final clinicopathologic response. MATERIALS AND METHODS: Institutional committees on clinical research and ethics approval and patient consent were obtained. Dynamic magnetic resonance (MR) examinations were performed in 25 women with primary breast cancer before treatment and after the first (n = 21) and second (n = 15) cycle of neoadjuvant chemotherapy. Kinetic parameters (transfer constant, leakage space, and rate constant) were derived for whole tumor regions of interest. Changes in histogram distributions of pixel data (median value and range) and MR imaging-derived size were correlated with final clinical and histologic response by using nonparametric methods. Receiver operating characteristic (ROC) analysis of tumor size and transfer constant changes were used to identify patients in whom no benefit was gained from chemotherapy. RESULTS: After the first cycle of treatment, 12 of 14 clinical responders showed decreases in tumor size, and six of seven nonresponders showed increases or no change in tumor size (P < .001). Transfer constant changes did not differ between responders and nonresponders for either clinical or pathologic assessments. After two cycles of treatment, there were tumor size increases in five of six nonresponders compared with decreases in eight of nine responders (P < .001). Reductions in transfer constant range were also observed in responders for both clinical and pathologic assessments (P = .008 and .02, respectively). No other kinetic parameter change predicted response. Size and transfer constant range were equally accurate for predicting the absence of pathologic response after two cycles of treatment (sensitivity, specificity, and area under ROC curve were 100%, 90%, and 0.93, respectively, for size and 100%, 75%, and 0.94, respectively, for transfer constant range). CONCLUSION: Reductions in MR imaging-determined size of the primary tumor best predict clinicopathologic response of breast cancer after one cycle of neoadjuvant chemotherapy. Transfer constant and size changes are equally sensitive in the identification of patients who would gain no clinical or pathologic benefit after two cycles of treatment.

Adult↗

PubMed2006

Noninvasive magnetic resonance spectroscopic pharmacodynamic markers of the choline kinase inhibitor MN58b in human carcinoma models.

MN58b is a novel anticancer drug that inhibits choline kinase, resulting in inhibition of phosphocholine synthesis. The aim of this work was to develop a noninvasive and robust pharmacodynamic biomarker for target inhibition and, potentially, tumor response following MN58b treatment. Human HT29 (colon) and MDA-MB-231 (breast) carcinoma cells were examined by proton (1H) and phosphorus (31P) magnetic resonance spectroscopy (MRS) before and after treatment with MN58b both in culture and in xenografts. An in vitro time course study of MN58b treatment was also carried out in MDA-MB-231 cells. In addition, enzymatic assays of choline kinase activity in cells were done. A decrease in phosphocholine and total choline levels (P < 0.05) was observed in vitro in both cell lines after MN58b treatment, whereas the inactive analogue ACG20b had no effect. In MDA-MB-231 cells, phosphocholine fell significantly as early as 4 hours following MN58b treatment, whereas a drop in cell number was observed at 48 hours. Significant correlation was also found between phosphocholine levels (measured by MRS) and choline kinase activities (r2 = 0.95, P = 0.0008) following MN58b treatment. Phosphomonoesters also decreased significantly (P < 0.05) in both HT29 and MDA-MB-231 xenografts with no significant changes in controls. 31P-MRS and 1H-MRS of tumor extracts showed a significant decrease in phosphocholine (P < or = 0.05). Inhibition of choline kinase by MN58b resulted in altered phospholipid metabolism both in cultured tumor cells and in vivo. Phosphocholine levels were found to correlate with choline kinase activities. The decrease in phosphocholine, total choline, and phosphomonoesters may have potential as noninvasive pharmacodynamic biomarkers for determining tumor response following treatment with choline kinase inhibitors.

Animals↗

PubMed2006

Factors influencing the accuracy of biomechanical breast models.

Recently it has been suggested that finite element methods could be used to predict breast deformations in a number of applications, including comparison of multimodality images, validation of image registration and image guided interventions. Unfortunately knowledge of the mechanical properties of breast tissues is limited. This study evaluated the accuracy with which biomechanical breast models based on finite element methods can predict the displacements of tissue within the breast in the practical clinical situation where the boundaries of the organ might be known reasonably accurately but there is some uncertainty on the mechanical properties of the tissue. For two datasets, we investigate the influence of tissue elasticity values, Poisson's ratios, boundary conditions, finite element solvers and mesh resolutions. Magnetic resonance images were acquired before and after compressing each volunteer's breast by about 20%. Surface displacement boundary conditions were derived from a three-dimensional nonrigid image registration. Six linear and three nonlinear elastic material models with and without skin were tested. These were compared to hyperelastic models. The accuracy of the models was evaluated by assessing the ability of the model to predict the location of 12 corresponding anatomical landmarks. The accuracy was most sensitive to the Poisson's ratio and the boundary condition. Best results were achieved for accurate boundary conditions, appropriate Poisson's ratios and models where fibroglandular tissue was at most four times stiffer than fatty tissue. These configurations reduced the mean (maximum) distance of the landmarks from 6.6 mm (12.4 mm) to 2.1 mm (3.4 mm) averaged over all experiments.

Algorithms↗

PubMed2006

Evaluation of a prospective scoring system designed for a multicenter breast MR imaging screening study.

PURPOSE: To evaluate prospectively the accuracy of a lesion classification system designed for use in a magnetic resonance (MR) imaging high-breast-cancer-risk screening study. MATERIALS AND METHODS: All participating patients provided written informed consent. Ethics committee approval was obtained. The results of 1541 contrast material-enhanced breast MR imaging examinations were analyzed; 1441 screening examinations were performed in 638 women aged 24-51 years at high risk for breast cancer, and 100 examinations were performed in 100 women aged 23-81 years. Lesion analysis was performed in 991 breasts, which were divided into design (491 breasts) and testing (500 breasts) sets. The reference standard was histologic analysis of biopsy samples, fine-needle aspiration cytology, or minimal follow-up of 24 months. The scoring system involved the use of five features: morphology (MOR), pattern of enhancement (POE), percentage of maximal focal enhancement (PMFE), maximal signal intensity-time ratio (MITR), and pattern of contrast material washout (POCW). The system was evaluated by means of (a) assessment of interreader agreement, as expressed in kappa statistics, for 315 breasts in which both readers analyzed the same lesion, (b) assessment of the diagnostic accuracy of the scored components with receiver operating characteristic curve analysis, and (c) logistic regression analysis to determine which components of the scoring system were critical to the final score. A new simplified scoring system developed with the design set was applied to the testing set. RESULTS: There was moderate reader agreement regarding overall lesion outcome (ie, malignant, suspicious, or benign) (kappa=0.58) and less agreement regarding the scored components. The area under the receiver operating characteristic curve (AUC) for the overall lesion score, 0.88, was higher than the AUC for any one component. The components MOR, POE, and POCW yielded the best overall result. PMFE and MITR did not contribute to diagnostic utility. Applying a simplified scoring system to the testing set yielded a nonsignificantly (P=.2) higher AUC than did applying the original scoring system (sensitivity, 84%; specificity, 86.0%). CONCLUSION: Good diagnostic accuracy can be achieved by using simple qualitative descriptors of lesion enhancement, including POCW. In the context of screening, quantitative enhancement parameters appear to be less useful for lesion characterization.

Adult↗

PubMed2006

Identification of magnetic resonance detectable metabolic changes associated with inhibition of phosphoinositide 3-kinase signaling in human breast cancer cells.

Phosphoinositide 3-kinase (PI3K) is an attractive target for novel mechanism-based anticancer treatment. We used magnetic resonance (MR) spectroscopy (MRS) to detect biomarkers of PI3K signaling inhibition in human breast cancer cells. MDA-MB-231, MCF-7, and Hs578T cells were treated with the prototype PI3K inhibitor LY294002, and the (31)P MR spectra of cell extracts were monitored. In every case, LY294002 treatment was associated with a significant decrease in phosphocholine levels by up to 2-fold (P < 0.05). In addition, a significant increase in glycerophosphocholine levels by up to 5-fold was also observed (P <or= 0.05), whereas the content of glycerophosphoethanolamine, when detectable, did not change significantly. Nucleotide triphosphate levels did not change significantly in MCF-7 and MDA-MB-231 cells but decreased by approximately 1.3-fold in Hs578T cells (P = 0.01). The changes in phosphocholine and glycerophosphocholine levels seen in cell extracts were also detectable in the (31)P MR spectra of intact MDA-MB-231 cells following exposure to LY294002. When treated with another PI3K inhibitor, wortmannin, MDA-MB-231 cells also showed a significant decrease in phosphocholine content by approximately 1.25-fold relative to the control (P < 0.05), whereas the levels of the remaining metabolites did not change significantly. Our results indicate that PI3K inhibition in human breast cancer cells by LY294002 and wortmannin is associated with a decrease in phosphocholine levels.

Androstadienes↗

PubMed2005

Inversion recovery measurements in the presence of radiation damping and implications for evaluating contrast agents in magnetic resonance.

Relaxation measurements performed at high magnetic field in magnetic resonance (MR) may be adversely affected by the influence of radiation damping in concentrated samples such as water. We consider how the measured value of T1 is affected by this phenomenon for a gadolinium-doped water sample and for an undoped water sample and consider the implications for evaluating contrast agents. A simple method involving the application of a pulsed field gradient to de-phase residual transverse components of the magnetization is shown to be an effective method for suppressing this effect. Given the central role that measurement of the T1 of water plays in the assessment of contrast agents as well as a host of other MR applications, care should always be employed when measuring and interpreting T1 measurements at high magnetic fields.

Brain↗

PubMed2005

A complete distortion correction for MR images: II. Rectification of static-field inhomogeneities by similarity-based profile mapping.

Radiotherapy treatment planning relies on the use of geometrically correct images. This paper presents a fully automatic tool for correcting MR images for the effects of B(0) inhomogeneities. The post-processing method is based on the gradient-reversal technique of Chang and Fitzpatrick (1992 IEEE Trans. Med. Imaging 11 319-29) which combines two identical images acquired with a forward- and a reversed read gradient. This paper demonstrates how maximization of mutual information for registration of forward and reverse read gradient images allows the elimination of user interaction for the correction. Image quality is preserved to a degree not reported previously.

Algorithms↗

PubMed2005

Identification of biliary metabolites of ifosfamide using 31P magnetic resonance spectroscopy and mass spectrometry.

Biliary excretion is a significant component in the metabolism of many drugs, but remains difficult to detect and characterise non-invasively. A previous publication recently described the detection of metabolites of ifosfamide in gall bladder in a guinea pig model using in vivo 1H-decoupled 31P 3-D magnetic resonance spectroscopic imaging and a clinical 1.5 T MR scanner.. Here high-resolution 31P magnetic resonance spectroscopy (MRS) of extracted bile identifies peaks as parent ifosfamide (1.19+/-1.47 mM; mean+/-sd), carboxyifosfamide (2.04+/-1.04 mM) and a major contribution from a previously unreported peak at 16.0 ppm (4.05+/-2.38 mM). The unknown resonance was identified using liquid chromatography-mass spectrometry (LCMS) as the glutathione conjugate of ifosfamide (MW=531). This was confirmed by analysing products from the reaction of glutathione with ifosfamide using LCMS and MRS. These results demonstrate how combined in vivo and analytical MRS, together with mass spectrometry, can help identify visceral routes of drug metabolism, thereby aiding understanding of +/-drug disposition and mechanisms of action and toxicity. In particular, the distribution of ifosfamide and its metabolites into bile may be related to oxazophosphorine-related cholecystitis reported in patients.

Animals↗

PubMed2005

Magnetic resonance spectroscopy monitoring of mitogen-activated protein kinase signaling inhibition.

Several mitogen-activated protein kinase (MAPK) signaling inhibitors are currently undergoing clinical trial as part of novel mechanism-based anticancer treatment strategies. This study was aimed at detecting biomarkers of MAPK signaling inhibition in human breast and colon carcinoma cells using magnetic resonance spectroscopy. We investigated the effect of the prototype MAPK kinase inhibitor U0126 on the (31)P-MR spectra of MDA-MB-231, MCF-7 and Hs578T breast, and HCT116 colon carcinoma cells. Treatment of MDA-MB-231 cells with 50 micromol/L U0126 for 2, 4, 8, 16, 24, 32, and 40 hours caused inhibition of extracellular signal-regulated kinases (ERK1/2) phosphorylation from 2 hours onwards. (31)P-MR spectra of extracted cells indicated that this was associated with a significant drop in phosphocholine levels to 78 +/- 8% at 8 hours, 74 +/- 8% at 16 hours, 66 +/- 7% at 24 hours, 71 +/- 10% at 32 hours, and 65 +/- 10% at 40 hours post-treatment. In contrast, the lower concentration of 10 micromol/L U0126 for 40 hours had no significant effect on either P-ERK1/ 2 or phosphocholine levels in MDA-MB-231 cells. Depletion of P-ERK1/2 in MCF-7 and Hs578T cells with 50 micromol/L U0126 also produced a drop in phosphocholine levels to 51 +/- 17% at 40 hours and 23 +/- 12% at 48 hours, respectively. Similarly, in HCT116 cells, inhibition with 30 micromol/L U0126 caused depletion of P-ERK1/2 and a decrease in phosphocholine levels to 80 +/- 9% at 16 hours and 61 +/- 4% at 24 hours post-treatment. The reduction in phosphocholine in MDA-MB-231 and HCT116 cells correlated positively with the drop in P-ERK1/2 levels. Our results show that MAPK signaling inhibition with U0126 is associated with a time-dependent decrease in cellular phosphocholine levels. Thus, phosphocholine has potential as a noninvasive pharmacodynamic marker for monitoring MAPK signaling blockade.

Biomarkers, Tumor↗

PubMed2005

A complete distortion correction for MR images: I. Gradient warp correction.

MR images are known to be distorted because of both gradient nonlinearity and imperfections in the B0 field, the latter caused either by an imperfect shim or sample-induced distortions. This paper describes in detail a method for correcting the gradient warp distortion, based on a direct field mapping using a custom-built phantom with three orthogonal grids of fluid-filled rods. The key advance of the current work over previous contributions is the large volume of the mapping phantom and the large distortions (>25 mm) corrected, making the method suitable for use with large field of view, extra-cranial images. Experimental measurements on the Siemens AS25 gradient set, as installed on a Siemens Vision scanner, are compared with a theoretical description of the gradient set, based on the manufacturer's spherical harmonic coefficients. It was found that over a volume of 320x200x340 mm3 distortions can be successfully mapped to within the voxel resolution of the raw imaging data, whilst outside this volume, correction is still good but some systematic errors are present. The phenomenon of through-plane distortion (also known as 'slice warp') is examined in detail, and the perturbation it causes to the measurements is quantified and corrected. At the very edges of the region of support provided by the phantom, through-plane distortion is extreme and only partially corrected by the present method. Solutions to this problem are discussed. Both phantom and patient data demonstrate the efficacy of the gradient warp correction.

Algorithms↗

PubMed2005

Localized COSY and DQF-COSY 1H-MRS sequences for investigating human tibial bone marrow in vivo and initial application to patients with acute leukemia.

PURPOSE: To develop 2D 1H-MRS measurement sequences for the evaluation of bone marrow lipids, and to assess these measurement sequences in healthy and diseased bone marrow. MATERIALS AND METHODS: Single-voxel localized variants of COSY and DQF-COSY 2D 1H-MRS sequences were developed for use at 1.5 T to investigate the biochemical composition of human bone marrow in vivo. The performance of each sequence was initially tested in vitro using lipid phantoms. An unsaturated lipid proton index was developed to interrogate the degree of unsaturation within the triacylglyceride (TAG) acyl chains. Localized 2D 1H-MRS data were obtained from the bone marrow of healthy controls (N = 6), patients presenting with acute leukemia (N = 6) and patients with acute leukemia in remission (N = 4). RESULTS: The COSY and DQF-COSY data recorded from all subject cohorts were similar, and the unsaturated lipid proton index did not reveal significant differences between patient groups. Variations in water content and measured relaxation times showed minor differences between the measurement groups. CONCLUSIONS: No significant differences were observed in the spectra obtained from bone marrow using the 2D 1H-MRS sequences. A novel unsaturated lipid proton index was developed.

Acute Disease↗

PubMed2005

Implementation and evaluation of CSI-localized J cross-polarization for detection of 31P magnetic resonance spectra in vivo.

Double resonance techniques such as INEPT (insensitive nuclei enhanced by polarization transfer) and JCP (J cross-polarization) have previously been applied in vitro to enhance the SNR of low-sensitivity nuclei and detect altered metabolism, for example, with 13C magnetic resonance spectroscopy (MRS), where the 1H-13C scalar couplings are of the order of 130 Hz. The aim of the present study was to investigate the potential advantage of using JCP for the detection of phosphomonoesters (PME) and phosphodiesters (PDE) with 31P MRS in vivo. These metabolites are involved in membrane metabolism and their concentration is altered in tumors and other pathologies. JCP has been implemented and compared with INEPT and pulse-and-acquire in vivo both in unlocalized and in localized spectra in order to select the optimum method for in vivo applications for PME and PDE detection. The results suggest that JCP can give up to 20% more signal in the PME region and up to 70% more signal in the PDE region, with 20 to 70% lower power deposition than INEPT. Such enhancement could be used to reduce the measurement times for equivalent signal-to-noise ratios. The JCP sequence is, however, slightly more sensitive than INEPT to RF field inhomogeneities, as predicted from theory.

Algorithms↗

PubMed2005

Quantitative assessment of the hepatic pharmacokinetics of the antimicrobial sitafloxacin in humans using in vivoF magnetic resonance spectroscopy.

AIMS: To measure hepatic concentrations of the fluorine-containing antimicrobial, sitafloxacin, using in vivo(19)F magnetic resonance spectroscopy (MRS). METHODS: Data were acquired from eight healthy subjects at 2, 5, 8 and 24 h following doses of 500 mg day(-1) for 5 days using a (1)H/(19)F surface coil in a 1.5T clinical MR system. Tissue water was used as a reference. RESULTS: Estimated liver concentrations at 2 h were 15.0 +/- 4.0 microg ml(-1) (mean +/- 95% CI), compared with 3.54 +/- 0.58 microg ml(-1) in plasma (n = 6), and fell below threshold concentrations (2 microg ml(-1)) by 24 h. CONCLUSIONS: (19)F MRS is able to detect and quantify sitafloxacin in the liver. There was no evidence for the hepatic retention of the drug.

Adolescent↗

PubMed2005

Reading protocol for dynamic contrast-enhanced MR images of the breast: sensitivity and specificity analysis.

PURPOSE: To prospectively determine sensitivity and specificity of breast magnetic resonance (MR) imaging in a screening and symptomatic population by using independent double reading, with histologic or cytologic results or a minimum 18-month follow-up as the standard. MATERIALS AND METHODS: Informed consent and ethical approval were obtained. Reader performance was analyzed in 44 radiologists at 18 centers from 1541 examinations, including 1441 screening examinations in 638 high-risk women aged 24-51 years (mean, 40.5 years) and 100 examinations in symptomatic women aged 23-81 years (mean, 49.2 years). A screening protocol of dynamic T1-weighted three-dimensional imaging and 0.2 mmol/kg gadolinium-based intravenous contrast agent was used. Logistic and Poisson regressions were used to analyze reader performance in relation to experience. Correlation between readers was determined with kappa statistics. Sensitivity and specificity were analyzed according to reader, field strength, machine type, and histologic results. RESULTS: The proportion of studies with lesions analyzed reduced significantly with reader experience (odds ratio, 0.84 per 6 months; P < .001), and number of regions per lesion analyzed also diminished (incidence rate ratio, 0.98 per 6 months; P = .047). The two readers for each study agreed 87% of the time, with a moderately good kappa statistic of 0.52 (95% confidence interval [CI]: 0.45, 0.58). By taking the reading with the highest score (most likely to be malignant) from each double-read study, sensitivity was 91% (95% CI: 83%, 96%) and specificity was 81% (95% CI: 79%, 83%). Single readings had 7% lower sensitivity (95% CI: 4%, 11%) and 7% higher specificity (95% CI: 6%, 7%). Sensitivity did not differ between MR imager manufacturers or between 1.0- and 1.5-T field strength, but there were significant differences in specificity for machine type (P = .001) and for field strength adjusted for manufacturer (P = .001). Specificity, but not sensitivity, was higher in women younger than 50 years (P = .02). CONCLUSION: Independent double reading by 44 radiologists blinded to mammography results showed sensitivity and specificity acceptable for screening; sensitivity was higher when two readings were used, at the cost of specificity. Interreader correlation was moderately good, and evidence of learning was seen. Equipment manufacturer, field strength, and age affected specificity but not sensitivity.

Adult↗

PubMed2004

Evaluation of radiological features for breast tumour classification in clinical screening with machine learning methods.

OBJECTIVE: In this work, methods utilizing supervised and unsupervised machine learning are applied to analyze radiologically derived morphological and calculated kinetic tumour features. The features are extracted from dynamic contrast enhanced magnetic resonance imaging (DCE-MRI) time-course data. MATERIAL: The DCE-MRI data of the female breast are obtained within the UK Multicenter Breast Screening Study. The group of patients imaged in this study is selected on the basis of an increased genetic risk for developing breast cancer. METHODS: The k-means clustering and self-organizing maps (SOM) are applied to analyze the signal structure in terms of visualization. We employ k-nearest neighbor classifiers (k-nn), support vector machines (SVM) and decision trees (DT) to classify features using a computer aided diagnosis (CAD) approach. RESULTS: Regarding the unsupervised techniques, clustering according to features indicating benign and malignant characteristics is observed to a limited extend. The supervised approaches classified the data with 74% accuracy (DT) and providing an area under the receiver-operator-characteristics (ROC) curve (AUC) of 0.88 (SVM). CONCLUSION: It was found that contour and wash-out type (WOT) features determined by the radiologists lead to the best SVM classification results. Although a fast signal uptake in early time-point measurements is an important feature for malignant/benign classification of tumours, our results indicate that the wash-out characteristics might be considered as important.

Artificial Intelligence↗