Optimal echo times for multi‐gradient echo‐based B0 field‐mapping

B₀ field maps are used ubiquitously in neuroimaging, in disciplines ranging from magnetic resonance spectroscopy to temperature mapping and susceptibility‐weighted imaging. Most B₀ maps are acquired using standard gradient‐echo–based vendor‐provided sequences, often comprised of two echoes spaced a few milliseconds apart. Herein, we analyze the optimal spacing of echo times, defined as those maximizing precision—minimizing the standard deviation—for a fixed total acquisition time. Field estimation is carried out using a weighted least squares estimator. The standard deviation is shown to be approximately inversely proportional to the total acquisition time, suggesting a law of diminishing returns, whereby substantial gains are obtained up to a certain point, with little improvement beyond that point. Validations are provided in a phantom and a group of volunteers. Multi‐gradient echo sequences are readily available on all manufacturer platforms, making our recommendations straightforward to implement on any modern scanner.     

Association between regional white and gray matter volume and ambiguity tolerance: Evidence from voxel‐based morphometry

The concept of tolerance of ambiguity (AT) is defined as the way in which an individual tends to perceive and deal with confusing, vague, and unclear situations. AT is generally considered as an important personality trait, but the neural mechanisms underlying individual differences in AT have never been investigated. Using voxel‐based morphometry and MSTAT‐II scale, we investigated the correlations between AT and regional white matter volume (rWMV) and regional gray matter volume (rGMV) in 351 young healthy subjects. We found AT to be positively correlated with rGMV in the dorsolateral prefrontal cortex (DLPFC), and negatively correlated with rGMV in the precuneus. These results indicate that increased rGMV in the left DLPFC may lead to characteristics of ambiguous stimuli consideration from multiple contexts and risk taking. Decreased rGMV in the left precuneus may be associated with a high tolerance for ambiguity, which attributes uncertainty to self‐related factors.     

Covarying structural alterations in laterality of the temporal lobe in schizophrenia: A case for source‐based laterality

The human brain is asymmetrically lateralized for certain functions (such as language processing) to regions in one hemisphere relative to the other. Asymmetries are measured with a laterality index (LI). However, traditional LI measures are limited by a lack of consensus on metrics used for its calculation. To address this limitation, source‐based laterality (SBL) leverages an independent component analysis for the identification of laterality‐specific alterations, identifying covarying components between hemispheres across subjects. SBL is successfully implemented with simulated data with inherent differences in laterality. SBL is then compared with a voxel‐wise analysis utilizing structural data from a sample of patients with schizophrenia and controls without schizophrenia. SBL group comparisons identified three distinct temporal regions and one cerebellar region with significantly altered laterality in patients with schizophrenia relative to controls. Previous work highlights reductions in laterality (ie, reduced left gray matter volume) in patients with schizophrenia compared with controls without schizophrenia. Results from this pilot SBL project are the first, to our knowledge, to identify covarying laterality differences within discrete temporal brain regions. The authors argue SBL provides a unique focus to detect covarying laterality differences in patients with schizophrenia, facilitating the discovery of laterality aspects undetected in previous work.     

T2 mapping with magnetization‐prepared 3D TSE based on a modified BIR‐4 T2 preparation

The purpose of this work was to investigate the performance of the modified BIR‐4 T₂ preparation for T₂ mapping and propose a method to remove T₂ quantification errors in the presence of large B₁ and B₀ offsets. The theoretical investigation of the magnetization evolution during the T₂ preparation in the presence of B₁ and B₀ offsets showed deviations from a mono‐exponential T₂ decay (two parameter fit). A three parameter fit was used to improve T₂ accuracy. Furthermore, a two parameter fit with an additional saturation preparation scan was proposed to improve T₂ accuracy and precision. These three fitting methods were compared based on simulations, phantom measurements and an in vivo healthy volunteer study of the neck musculature using a 3D TSE readout. The results based upon the pure two parameter fit overestimated T₂ in regions with high B₀ offsets (up to 40% in phantoms). The three parameter fit T₂ values were robust to B₀ offsets but with higher standard deviation (up to 40% in simulations). The two parameter fit with the saturation preparation yielded high robustness towards B₀ offsets with a noise performance comparable to that of the two parameter fit. In the volunteer study the T₂ values obtained by the pure two parameter fit showed a dependence on the field inhomogeneities, whereas the T₂ values from the proposed fitting approach were shown to be insensitive to B₀ offsets. The proposed method enabled accurate and precise T₂ mapping in the presence of large B₁ and B₀ offsets.     

Quantitative susceptibility mapping and 23Na imaging‐based in vitro characterization of blood clotting kinetics

Blood clotting is a fundamental biochemical process in post‐hemorrhagic hemostasis. Although the varying appearance of coagulating blood in T₁‐ and T₂‐weighted images is widely used to qualitatively determine bleeding age, the technique permits only a rough discrimination of coagulation stages, and it remains difficult to distinguish acute and chronic hemorrhagic stages because of low T₁‐ and T₂‐weighted signal intensities in both instances. To investigate new biomedical parameters for magnetic resonance imaging‐based characterization of blood clotting kinetics, sodium imaging and quantitative susceptibility mapping (QSM) were compared with conventional T₁‐ and T₂‐weighted imaging, as well as with biochemical hemolysis parameters. For this purpose, a blood‐filled spherical agar phantom was investigated daily for 14 days, as well as after 24 days at 7 T after initial preparation with fresh blood. T₁‐ and T₂‐weighted sequences, a three‐dimensional (3D) gradient echo sequence and a density‐adapted 3D radial projection reconstruction pulse sequence for ²³Na imaging were applied. For hemolysis estimations, free hemoglobin and free potassium concentrations were measured photometrically and with the direct ion‐selective electrode method, respectively, in separate heparinized whole‐blood samples along the same timeline. Initial mean susceptibility was low (0.154 ± 0.020 ppm) and increased steadily during the course of coagulation to reach up to 0.570 ± 0.165 ppm. The highest total sodium (NaT) values (1.02 ± 0.06 arbitrary units) in the clot were observed initially, dropped to 0.69 ± 0.13 arbitrary units after one day and increased again to initial values. Compartmentalized sodium (NaS) showed a similar signal evolution, and the NaS/NaT ratio steadily increased over clot evolution. QSM depicts clot evolution in vitro as a process associated with hemoglobin accumulation and transformation, and enables the differentiation of the acute and chronic coagulation stages. Sodium imaging visualizes clotting independent of susceptibility and seems to correspond to clot integrity. A combination of QSM and sodium imaging may enhance the characterization of hemorrhage.     

Comparison of spoiled gradient echo and steady‐state free‐precession imaging for native myocardial T1 mapping using the slice‐interleaved T1 mapping (STONE) sequence

Cardiac T₁ mapping allows non‐invasive imaging of interstitial diffuse fibrosis. Myocardial T₁ is commonly calculated by voxel‐wise fitting of the images acquired using balanced steady‐state free precession (SSFP) after an inversion pulse. However, SSFP imaging is sensitive to B₁ and B₀ imperfection, which may result in additional artifacts. A gradient echo (GRE) imaging sequence has been used for myocardial T₁ mapping; however, its use has been limited to higher magnetic field to compensate for the lower signal‐to‐noise ratio (SNR) of GRE versus SSFP imaging. A slice‐interleaved T₁ mapping (STONE) sequence with SSFP readout (STONE–SSFP) has been recently proposed for native myocardial T₁ mapping, which allows longer recovery of magnetization (>8 R–R) after each inversion pulse. In this study, we hypothesize that a longer recovery allows higher SNR and enables native myocardial T₁ mapping using STONE with GRE imaging readout (STONE–GRE) at 1.5T. Numerical simulations and phantom and in vivo imaging were performed to compare the performance of STONE–GRE and STONE–SSFP for native myocardial T₁ mapping at 1.5T. In numerical simulations, STONE–SSFP shows sensitivity to both T₂ and off resonance. Despite the insensitivity of GRE imaging to T₂, STONE–GRE remains sensitive to T₂ due to the dependence of the inversion pulse performance on T₂. In the phantom study, STONE–GRE had inferior accuracy and precision and similar repeatability as compared with STONE–SSFP. In in vivo studies, STONE–GRE and STONE–SSFP had similar myocardial native T₁ times, precisions, repeatabilities and subjective T₁ map qualities. Despite the lower SNR of the GRE imaging readout compared with SSFP, STONE–GRE provides similar native myocardial T₁ measurements, precision, repeatability, and subjective image quality when compared with STONE–SSFP at 1.5T.     

Integrated Laplacian‐based phase unwrapping and background phase removal for quantitative susceptibility mapping

Quantitative susceptibility mapping (QSM) is a recently developed MRI technique that provides a quantitative measure of tissue magnetic susceptibility. To compute tissue magnetic susceptibilities based on gradient echoes, QSM requires reliable unwrapping of the measured phase images and removal of contributions caused by background susceptibilities. Typically, the two steps are performed separately. Here, we present a method that simultaneously performs phase unwrapping and HARmonic (background) PhasE REmovaL using the LAplacian operator (HARPERELLA). Both numerical simulations and in vivo human brain images show that HARPERELLA effectively removes both phase wraps and background phase, whilst preserving all low spatial frequency components originating from brain tissues. When compared with other QSM phase preprocessing techniques, such as path‐based phase unwrapping followed by background phase removal, HARPERELLA preserves the tissue phase signal in gray matter, white matter and cerebrospinal fluid with excellent robustness, providing a convenient and accurate solution for QSM. The proposed algorithm is provided, together with QSM and susceptibility tensor imaging (STI) tools, in a shared software package named ‘STI Suite’. Copyright © 2013 John Wiley & Sons, Ltd.     

Simultaneous multi‐slice spin‐ and gradient‐echo dynamic susceptibility‐contrast perfusion‐weighted MRI of gliomas

Although combined spin‐ and gradient‐echo (SAGE) dynamic susceptibility‐contrast (DSC) MRI can provide perfusion quantification that is sensitive to both macrovessels and microvessels while correcting for T₁‐shortening effects, spatial coverage is often limited in order to maintain a high temporal resolution for DSC quantification. In this work, we combined a SAGE echo‐planar imaging (EPI) sequence with simultaneous multi‐slice (SMS) excitation and blipped controlled aliasing in parallel imaging (blipped CAIPI) at 3 T to achieve both high temporal resolution and whole brain coverage. Two protocols using this sequence with multi‐band (MB) acceleration factors of 2 and 3 were evaluated in 20 patients with treated gliomas to determine the optimal scan parameters for clinical use. ΔR₂*(t) and ΔR₂(t) curves were derived to calculate dynamic signal‐to‐noise ratio (dSNR), ΔR₂*‐ and ΔR₂‐based relative cerebral blood volume (rCBV), and mean vessel diameter (mVD) for each voxel. The resulting SAGE DSC images acquired using MB acceleration of 3 versus 2 appeared visually similar in terms of image distortion and contrast. The difference in the mean dSNR from normal‐appearing white matter (NAWM) and that in the mean dSNR between NAWM and normal‐appearing gray matter were not statistically significant between the two protocols. ΔR₂*‐ and ΔR₂‐rCBV maps and mVD maps provided unique contrast and spatial heterogeneity within tumors.     

MRI of carriers of the apolipoprotein E e4 allele–evidence for structural differences in normal‐appearing brain tissue in e4+ relative to e4– young adults

Apolipoprotein E is a protein involved in cholesterol and lipid transport. The gene coding for this protein has three different alleles: e2, e3 and e4. The e4 allele is recognised as a significant risk factor for the development of Alzheimer's disease in later life. Paradoxically, behavioural and functional evidence has demonstrated that the e4 allele may confer a cognitive advantage to the carrier in youth. In this article, a range of sophisticated and novel structural imaging techniques were used to identify subtle differences in the brain tissue of groups of young e4 and homozygous e3 carriers that might support this paradox. Using voxel‐based morphometry of high‐resolution structural MR images, we identified a higher white matter volume ratio in e4 relative to homozygous e3 carriers. Furthermore, diffusion tensor imaging and tract‐based spatial statistics studies identified increases in axial diffusivity and mode of anisotropy in carriers of the e4 allele. In addition, quantitative magnetisation transfer data were analysed using tract‐based spatial statistics. Evidence of a trend towards an increased transverse relaxation time of the bound proton pool was detected in e4 carriers, indicative of altered white matter composition. These changes were found to correlate with indices of cognitive performance across the two groups, supporting the notion that such subtle differences in white matter integrity may confer neural advantages that contribute to cognitive outcomes and, potentially, to performance differences, such as observed here in a test of verbal fluency and reported previously by other researchers. Copyright © 2013 John Wiley & Sons, Ltd.     

Simultaneous MR quantification of hepatic fat content,fatty acid composition,transverse relaxation time and magnetic susceptibility for the diagnosis of non‐alcoholic steatohepatitis

Non‐alcoholic steatohepatitis (NASH) is characterized at histology by steatosis, hepatocyte ballooning and inflammatory infiltrates, with or without fibrosis. Although diamagnetic material in fibrosis and inflammation can be detected with quantitative susceptibility imaging, fatty acid composition changes in NASH relative to simple steatosis have also been reported. Therefore, our aim was to develop a single magnetic resonance (MR) acquisition and post‐processing scheme for the diagnosis of steatohepatitis by the simultaneous quantification of hepatic fat content, fatty acid composition, T ₂* transverse relaxation time and magnetic susceptibility in patients with non‐alcoholic fatty liver disease. MR acquisition was performed at 3.0 T using a three‐dimensional, multi‐echo, spoiled gradient echo sequence. Phase images were unwrapped to compute the B ₀ field inhomogeneity (ΔB ₀) map. The ΔB ₀‐demodulated real part images were used for fat–water separation, T ₂* and fatty acid composition quantification. The external and internal fields were separated with the projection onto dipole field method. Susceptibility maps were obtained after dipole inversion from the internal field map with single‐orientation Bayesian regularization including spatial priors. Method validation was performed in 32 patients with biopsy‐proven, non‐alcoholic fatty liver disease from which 12 had simple steatosis and 20 NASH. Liver fat fraction and T ₂* did not change significantly between patients with simple steatosis and NASH. In contrast, the saturated fatty acid fraction increased in patients with NASH relative to patients with simple steatosis (48 ± 2% versus 44 ± 4%; p  < 0.05) and the magnetic susceptibility decreased (?0.30 ± 0.27 ppm versus 0.10 ± 0.14 ppm; p  < 0.001). The area under the receiver operating characteristic curve for magnetic susceptibility as NASH marker was 0.91 (95% CI: 0.79–1.0). Simultaneous MR quantification of fat content, fatty acid composition, T ₂* and magnetic susceptibility is feasible in the liver. Our preliminary results suggest that quantitative susceptibility imaging has a high diagnostic performance for the diagnosis of NASH.     

Multi‐echo susceptibility‐weighted imaging and histology of open‐field blast‐induced traumatic brain injury in a rat model

Blast‐induced traumatic brain injury is on the rise, predominantly as a result of the use of improvised explosive devices, resulting in undesirable neuropsychological dysfunctions, as demonstrated in both animals and humans. This study investigated the effect of open‐field blast injury on the rat brain using multi‐echo, susceptibility‐weighted imaging (SWI). Multi‐echo SWI provided phase maps with better signal‐to‐noise ratio (SNR) and contrast‐to‐noise ratio (CNR), making it a sensitive technique for brain injury. Male Sprague‐Dawley rats were subjected to a survivable blast of 180 kPa. The visibility of blood vessels of varying sizes improved with multi‐echo SWI. Reduced signal intensity from major vessels post‐blast indicates increased deoxyhaemoglobin. Relative cerebral blood flow was computed from filtered phase SWI images using inferred changes in oxygen saturation from major blood vessels. Cerebral blood flow decreased significantly at day 3 and day 5 post‐blast compared with that pre‐blast. This was substantiated by the upregulation of β‐amyloid precursor protein (β‐APP), a marker of ischaemia, in the neuronal perikaya of the cerebral cortex, as observed by immunofluorescence, and in the cortical tissue by western blot analysis. Our findings indicate the presence of brain ischaemia in post‐blast acute phase of injury with possible recovery subsequently. Our results from cerebrovascular imaging, histology and staining provide an insight into the ischaemic state of the brain post‐blast and may be useful for prognosis and outcome. Copyright © 2015 John Wiley & Sons, Ltd.     

Diffusion‐prepared stimulated‐echo turbo spin echo (DPsti‐TSE): An eddy current‐insensitive sequence for three‐dimensional high‐resolution and undistorted diffusion‐weighted imaging

In this study, we present a new three‐dimensional (3D), diffusion‐prepared turbo spin echo sequence based on a stimulated‐echo read‐out (DPsti‐TSE) enabling high‐resolution and undistorted diffusion‐weighted imaging (DWI). A dephasing gradient in the diffusion preparation module and rephasing gradients in the turbo spin echo module create stimulated echoes, which prevent signal loss caused by eddy currents. Near to perfect agreement of apparent diffusion coefficient (ADC) values between DPsti‐TSE and diffusion‐weighted echo planar imaging (DW‐EPI) was demonstrated in both phantom transient signal experiments and phantom imaging experiments. High‐resolution and undistorted DPsti‐TSE was demonstrated in vivo in prostate and carotid vessel wall. 3D whole‐prostate DWI was achieved with four b values in only 6 min. Undistorted ADC maps of the prostate peripheral zone were obtained at low and high imaging resolutions with no change in mean ADC values [(1.60 ± 0.10) × 10^?3 versus (1.60 ± 0.02) × 10^?3 mm²/s]. High‐resolution 3D DWI of the carotid vessel wall was achieved in 12 min, with consistent ADC values [(1.40 ± 0.23) × 10^?3 mm²/s] across different subjects, as well as slice locations through the imaging volume. This study shows that DPsti‐TSE can serve as a robust 3D diffusion‐weighted sequence and is an attractive alternative to the traditional two‐dimensional DW‐EPI approaches.     

Quantitative susceptibility mapping of the spine using in‐phase echoes to initialize inhomogeneous field and R2* for the nonconvex optimization problem of fat‐water separation

Quantitative susceptibility mapping (QSM) of human spinal vertebrae from a multi‐echo gradient‐echo (GRE) sequence is challenging, because comparable amounts of fat and water in the vertebrae make it difficult to solve the nonconvex optimization problem of fat‐water separation (R2*‐IDEAL) for estimating the magnetic field induced by tissue susceptibility. We present an in‐phase (IP) echo initialization of R2*‐IDEAL for QSM in the spinal vertebrae. Ten healthy human subjects were recruited for spine MRI. A 3D multi‐echo GRE sequence was implemented to acquire out‐phase and IP echoes. For the IP method, the R2* and field maps estimated by separately fitting the magnitude and phase of IP echoes were used to initialize gradient search R2*‐IDEAL to obtain final R2*, field, water, and fat maps, and the final field map was used to generate QSM. The IP method was compared with the existing Zero method (initializing the field to zero), VARPRO‐GC (variable projection using graphcuts but still initializing the field to zero), and SPURS (simultaneous phase unwrapping and removal of chemical shift using graphcuts for initialization) on both simulation and in vivo data. The single peak fat model was also compared with the multi‐peak fat model. There was no substantial difference on QSM between the single peak and multi‐peak fat models, but there were marked differences among different initialization methods. The simulations demonstrated that IP provided the lowest error in the field map. Compared to Zero, VARPRO‐GC and SPURS, the proposed IP method provided substantially improved spine QSM in all 10 subjects.     

Comparison of reproducibility of single voxel spectroscopy and whole‐brain magnetic resonance spectroscopy imaging at 3T

To date, single voxel spectroscopy (SVS) is the most commonly used MRS technique. SVS is relatively easy to use and provides automated and immediate access to the resulting spectra. However, it is also limited in spatial coverage. A new and very promising MRS technique allows for whole‐brain MR spectroscopic imaging (WB‐MRSI) with much improved spatial resolution. Establishing the reproducibility of data obtained using SVS and WB‐MRSI is an important first step for using these techniques to evaluate longitudinal changes in metabolite concentration. The purpose of this study was to assess and directly compare the reproducibility of metabolite quantification at 3T using SVS and WB‐MRSI in ‘hand‐knob’ areas of motor cortices and hippocampi in healthy volunteers. Ten healthy adults were scanned using both SVS and WB‐MRSI on three occasions one week apart. N‐acetyl aspartate (NAA), creatine (Cr), choline (Cho) and myo‐inositol (mI) were quantified using SVS and WB‐MRSI with reference to both Cr and H₂O. The reproducibility of each technique was evaluated using the coefficient of variation (CV), and the correspondence between the two techniques was assessed using Pearson correlation analysis. The measured mean (range) intra‐subject CVs for SVS were 5.90 (2.65‐10.66)% for metabolites (i.e. NAA, Cho, mI) relative to Cr, and 8.46 (4.21‐21.07)% for metabolites (NAA, Cr, Cho, mI) relative to H₂O. The mean (range) CVs for WB‐MRSI were 7.56 (2.78‐11.41)% for metabolites relative to Cr, and 7.79 (4.57‐14.11)% for metabolites relative to H₂O. Significant positive correlations were observed between metabolites quantified using SVS and WB‐MRSI techniques when the Cr but not H₂O reference was used. The results demonstrate that reproducibilities of SVS and WB‐MRSI are similar for quantifying the four major metabolites (NAA, Cr, Cho, mI); both SVS and WB‐MRSI exhibited good reproducibility. Our findings add reference information for choosing the appropriate ¹H‐MRS technique in future studies.     

Quantitative susceptibility mapping of kidney inflammation and fibrosis in type 1 angiotensin receptor‐deficient mice

Disruption of the regulatory role of the kidneys leads to diverse renal pathologies; one major hallmark is inflammation and fibrosis. Conventional magnitude MRI has been used to study renal pathologies; however, the quantification or even detection of focal lesions caused by inflammation and fibrosis is challenging. We propose that quantitative susceptibility mapping (QSM) may be particularly sensitive for the identification of inflammation and fibrosis. In this study, we applied QSM in a mouse model deficient for angiotensin receptor type 1 (AT₁). This model is known for graded pathologies, including focal interstitial fibrosis, cortical inflammation, glomerulocysts and inner medullary hypoplasia. We acquired high‐resolution MRI on kidneys from AT₁‐deficient mice that were perfusion fixed with contrast agent. Two MR sequences were used (three‐dimensional spin echo and gradient echo) to produce three image contrasts: T₁, T₂* (magnitude) and QSM. T₁ and T₂* (magnitude) images were acquired to segment major renal structures and to provide landmarks for the focal lesions of inflammation and fibrosis in the three‐dimensional space. The volumes of major renal structures were measured to determine the relationship of the volumes to the degree of renal abnormalities and magnetic susceptibility values. Focal lesions were segmented from QSM images and were found to be closely associated with the major vessels. Susceptibilities were relatively more paramagnetic in wild‐type mice: 1.46 ± 0.36 in the cortex, 2.14 ± 0.94 in the outer medulla and 2.10 ± 2.80 in the inner medulla (10^–2 ppm). Susceptibilities were more diamagnetic in knockout mice: –7.68 ± 4.22 in the cortex, –11.46 ± 2.13 in the outer medulla and –7.57 ± 5.58 in the inner medulla (10^–2 ppm). This result was consistent with the increase in diamagnetic content, e.g. proteins and lipids, associated with inflammation and fibrosis. Focal lesions were validated with conventional histology. QSM was very sensitive in detecting pathology caused by small focal inflammation and fibrosis. QSM offers a new MR contrast mechanism to study this common disease marker in the kidney. Copyright © 2013 John Wiley & Sons, Ltd.     

Nuclear morphometry and texture analysis of B‐cell non‐Hodgkin lymphoma: Utility in subclassification on cytosmears

Non‐Hodgkin lymphoma (NHL) is a heterogeneous group of lymphoid neoplasms and accurate subclassification is an essential prerequisite for proper management of patients. This study was aimed at evaluating the utility of nuclear morphometry and textural features on cytology smears to classify the cases of NHL on aspiration cytology. Fine needle aspiration smears of 50 cases of B‐cell NHL were included. Various morphometric and texture parameters were obtained by manually tracing the nuclei on digitized images in each case and discriminant analysis performed using various features taken individually as well as all together. The percentage of cells correctly classified to a particular NHL subtype using the discriminant functions so obtained was noted. Our results show that discriminant analysis done on size parameters could correctly classify a greater number of cells than on shape parameters (36.4% vs. 21.2%, respectively). Texture parameters based on single pixel values (first order texture) were inferior (42.8%) to those based on pair of pixels (58.7%) in subtyping of cells. Discriminant analysis based on color parameters was more effective (61.9%) as compared to rest of the morphometric and textural parameters. Using all the morphometric and textural parameters together, 83.3% of cells could be correctly classified to a particular NHL subtype. The present study, perhaps the first study of detailed morphometric analysis on cytosmears, shows that satisfactory classification of NHL on aspiration cytology is possible using nuclear morphometry and textural parameters considered together. These results are promising for further studies on this subject and development of automated cytodiagnosis. Diagn. Cytopathol. 2010. © 2009 Wiley‐Liss, Inc.