Dosimetric Performance Evaluation of MLC-based and Cone-based 3D Spatially Fractionated LATTICE Radiotherapy.

This study aims to dosimetrically compare multi-leaf collimator (MLC)-based and cone-based 3D LATTICE radiotherapy (LRT) plans. Valley-peak ratios were evaluated using seven different 3D LATTICE designs. Target volumes of 8 cm and 12 cm were defined on the RANDO phantom. Valley-peak dose patterns were obtained by creating high-dose vertices in the target volumes. By changing the vertex diameter, vertices separation, and volume ratio, seven different LATTICE designs were generated. Treatment plans were implemented using CyberKnife and Varian RapidArc. Thermoluminescent dosimeter (TLD), EBT3 films, and electronic portal-imaging device (EPID) were employed for dosimetric treatment verification, and measured doses were compared to calculated doses. By changing the vertex diameter and vertices separation, the valley-peak ratio was exhibited little difference between the two systems. By changing the vertex diameter and volume ratio, the valley-peak ratio was observed nearly the same for the two systems. The film, TLD, and EPID dosimetry showed good agreement between the calculated and measured doses. Based on the results, we concluded that although smaller valley-peak ratios were obtained with cone-based plans, the dose-volume histograms were comparable in both systems. Also, when we evaluated the treatment duration, the MLC-based plans were more appropriate to apply the treatment in a single fraction.

Planning and dosimetric evaluation of three total body irradiation techniques: Standard SSD VMAT, Extended SSD VMAT and Extended SSD Field-in-Field.

The aim of this study was to dosimetrically compare three total body irradiation (TBI) techniques which can be delivered by a standard linear accelerator, and to deduce which one is preferable. Specifically, Extended Source to Surface Distance (SSD) Field-in-Field (FiF), Extended SSD Volumetric Modulated Arc Therapy (VMAT), and Standard SSD VMAT TBI techniques were dosimetrically evaluated. Percent depth dose and dose profile measurements were made under treatment conditions for each specified technique. After having generated treatment plans with a treatment planning system (TPS), dose homogeneity and critical organ doses were investigated on a Rando phantom using radiochromic films and optically stimulated luminescence dosimeters (OSLDs). TBI dose of 12 Gy in six fractions was prescribed for each technique. The gamma index (5%/5 mm) was used for the analysis of radiochromic films. Passing rates for Extended SSD FiF, Extended SSD VMAT and Standard SSD VMAT techniques were found to be 90%, 87% and 94%, respectively. OSLD measurements were within ± 5% agreement with TPS calculations for the first two techniques whereas the agreement was found to be within ± 3% for the Standard SSD VMAT technique. TPS calculations demonstrated that mean lung doses in the first two techniques were around 8.5 Gy while it was kept around 7 Gy in Standard SSD VMAT. It is concluded that Standard SSD VMAT is superior in sparing the lung tissue while all three TBI techniques are feasible in clinical practice with acceptable dose homogeneity. In the absence of VMAT-based treatment planning, Extended SSD FiF would be a reasonable choice compared to other conventional techniques.

Determination of inflection points of CyberKnife dose profiles within acceptability criteria of deviations in measurements.

AIM: The aim of this study was to determine the Inflection Points (IPs) of flattening filter free (FFF) CyberKnife dose profiles for cone-based streotactic radiotherapy. In addition, dosimetric field sizes were determined. BACKGROUND: The increased need for treatment in the early stages of cancer necessitated the treatment of smaller tumors. However, efforts in that direction required the modeling accuracy of the beam. Removal of the flattening filter (FF) from the path of x-ray beam has provided the solution to those efforts, but required a different normalization approach for the beam to ensure the delivery of the dose accurately. As a solution, researchers proposed a normalization factor based on IPs. MATERIALS AND METHODS: Measurements using microDiamond (PTW 60019), Diode SRS (PTW 60018) and Monte Carlo (MC) calculations of dose profiles were completed at SAD 80 cm and 5 cm depth for 15-60 mm cones. Performance analysis of detectors with respect to MC calculation was carried out. Gamma evaluation method was used to determine achievable acceptability criteria for FFF CyberKnife beams. RESULTS: Acceptability within (3%-0.5 mm) was found to be anachievable criterion for all dose profile measurements of the cone beams used in this study. To determine the IP, the first and second derivatives of the dose profile were determined via the cubic spline interpolation technique. CONCLUSION: Derivatives of the interpolated profiles showed that locations of IPs and 50% isodose points coincide.

Evaluation of NanoDot Optically Stimulated Luminescence Dosimeter for Cone-shaped Small-field Dosimetry of Cyberknife Stereotactic Radiosurgery Unit: A Monte Carlo Simulation and Dosimetric Verification Study.

AIM: The aim of this study was to investigate the adequacy of nanoDot optically stimulated luminescence (OSL) dosimeter for small field dosimetry before its in vivo applications in CyberKnife SRS unit. MATERIALS AND METHODS: A PTW 60018 SRS Diode, 60019 microDiamond, and Gafchromic EBT3 films were used along with a nanoDot carbon-doped aluminum oxide OSL dosimeter to collect and compare beam data. In addition, the EGSnrc/BEAMnrc code was employed to simulate 6-MV photon beams of CyberKnife SRS system. RESULTS: All detectors showed good consistency with each other in output factor measurements for cone sizes of 15 mm or more. The differences were maintained within 3% for these cones. However, OSL output factors showed higher discrepancies compared to those of other detectors for smaller cones wherein the difference reached nearly 40% for cone size of 5 mm. Depending on the performance of OSL dosimeter in terms of output factors, percentage depth doses (PDDs) were only measured for cones equal to or larger than 15 mm. The differences in PDD measurements were within 5% for depths in the range of 5-200 mm. CONCLUSION: Its low reliable readings for cones smaller than 15 mm should be considered before its in vivo applications of Cyberknife system.

Performance evaluation of the X-sight spine tracking system for abdominal tumors distal to spine: A 2D dosimetric analysis.

The X-sight spine tracking system was integrated with Cyberknife unit to deliver radiosurgery treatments for spinal tumors without fiducial placement. However the tracking system can also be used for the treatment of abdominal tumors located in a certain distance from the spine. The aim of our study is to evaluate the tracking performance of the X-sight spine system for abdominal tumors distal to spine based on the 3 factors: tumor distance from the reference vertebra, the angle of tumor with the vertebra, and the amplitude of tumor motion due to respiration. An experimental setup was designed mainly with ovine lumbar vertebrae and the BrainLab ExacTrac gating phantom. Planning Target Volume (PTV) structures were delineated at different vertical distances from the reference vertebra. The dosimetric measurements were taken with GafChromic EBT3 film placed between slab phantoms so that they corresponded to centers of the target volumes. Dosimetric comparisons were performed based on dose-volume parameters and the gamma analysis. The measurements were then repeated for the same experimental conditions by using the Synchrony system to compare tracking performances. Using the X-sight system, percentage differences between the dose-volume parameters of the Treatment Planning System (TPS) calculations and the EBT3 film readings went up to 12% for the motion amplitude of 8 mm. The differences decreased with small motions while angles and vertical distances of the lesion locations did not induce major changes in dose discrepancies. Percentages of pixels passing gamma analysis were found to be below the acceptance threshold of 95%. Using the Synchrony system, the measured dose distributions had more similar patterns with those of the TPS system such that the percentage differences in the dose parameters were less than 4% and the gamma passing rates were found to be higher than 95%. Our results showed that the X-sight spine system should not be chosen for tracking abdominal tumors distal to the spine or osseous structures because of the effect of diaphragmatic motion on entire abdominal region. The fiducial-based Synchrony tracking system can be preferred for these tumors.

Fully automated blink detection for uveal melanoma radiotherapy.

Uveal melanoma is a widespread neoplasm in the eye that cause vision defects. The disease may result in loss of eye, even loss of life with a risk of metastasis to other tissues. There have been proposed various treatment methods to overcome the disease including enucleation and radiotherapy. Radiotherapy treatment is certainly preferable since it avoids the loss of eye. The treatment is applied with radioactive beam directed to the tumor, and thus the eye must be kept steady usually with local anesthesia. However, anesthesia has many health risks; therefore it is preferable to perform this operation without it. This paper presents an approach where a camera will be recording the eye and the radiotherapy will immediately be halted when the eye is closed to avoid unnecessary beams directed to healthy regions of the eye. System is based on the classification of the acquired image frames as "eye open" and "eye closed". The speed of the algorithm is aimed as close as possible to real time so that it can respond instantly to any change in present eye state. A total of 480 video frames, 120 frames for each open and closed eyes in bright and dark lighting conditions separately, were used for experimentation. Test results indicate that developed features are successful in terms of being very fast providing no classification error.

Investigating the surface dose contribution of intrafractional kV imaging in CyberKnife-based stereotactic radiosurgery.

CyberKnife treatment consists of hundreds of noncoplanar beams and numerous intrafractional images that can be taken during a single treatment fraction; thus, doses because of imaging should be considered in this technique. The aim of this study is to investigate the in-field and out-of-field surface doses induced from kV imaging system during stereotactic radiosurgery (SRS) treatment. The imaging-induced surface doses were measured at the center of the imaging field and within ±15-cm distance from the center in both craniocaudal and lateral directions. TLD100H thermoluminescence dosimeters and EBT2 gafchromic films were used to take the measurements at the locations of 0, ±5, ±10, and ±15 cm in the 2 orthogonal directions on abdominal region of a Rando phantom. The surface dose contributions of imaging system for the 4 most commonly used energy options of 90, 100, 110, and 120 kVp with 3 mAs options of 10, 30, and 90 mAs were measured and compared. Imaging dose values have a positive correlation with both parameters of energy and mAs. The energy options of 100, 110, and 120 kVp, in average, induced 60%, 101%, and 141% more doses per mAs than 90 kVp energy in the imaging field center. A threefold increase in mAs values, i.e., from 10 mAs to 30 mAs and from 30 mAs to 90 mAs, caused higher dose in field center with a factor of 2.53 ± 0.08 when the energy value was kept constant. The in-field dose distributions within ±10 cm in both directions showed a flat pattern with a standard deviation lower than 5%, whereas the out-of-field doses at ±15-cm distance from the field center suddenly dropped to almost half of the central doses. Although a single imaging attempt causes a very low dose compared with the therapeutic dose level, one should be aware of the cumulative surface dose increase with higher imaging number. Proper patient setup, fiducial usage, and reduction of both the mAs values and the imaging numbers should be, therefore, considered to keep the cumulative surface dose in a lower level.

Evaluation of MLC leaf positioning accuracy for static and dynamic IMRT treatments using DAVID in vivo dosimetric system.

Accuracy and precision of leaf positioning in multileaf collimators (MLCs) are significant factors for the accuracy of IMRT treatments. This study aimed to inves-tigate the accuracy and repeatability of the MLC leaf positioning via the DAVID invivo dosimetric system for dynamic and static MLC systems. The DAVID system was designed as multiwire transmission ionization chamber which is placed in accessory holder of linear accelerators. Each wire of DAVID system corresponds to a MLC leaf-pair to verify the leaf positioning accuracy during IMRT treatment and QA. In this study, verifications of IMRT plans of five head and neck (H&N) and five prostate patients treated in a Varian DHX linear accelerator with 80-leaf MLC were performed using DAVID system. Before DAVID-based dosimetry, Electronics Portal Imaging Device (EPID) and PTW 2D ARRAY dosimetry system were used for 2D verification of each plan. The measurements taken by DAVID system in the first day of the treatments were used as reference for the following measurements taken over the next four weeks. The deviations in leaf positioning were evaluated by "Total Deviation (TD)" parameter calculated by DAVID software. The delivered IMRT plans were originally prepared using dynamic MLC method. The same plans were subsequently calculated based on static MLC method with three different intensity levels of five (IL5), 10 (IL10) and 20 (IL20) in order to compare the performances of MLC leaf positioning repeatability for dynamic and static IMRT plans. The leaf positioning accuracy is also evaluated by analyzing DynaLog files based on error histograms and root mean square (RMS) errors of leaf pairs' positions. Moreover, a correlation analysis between simultaneously taken DAVID and EPID measurements and DynaLog file recordings was subsequently performed. In the analysis of DAVID outputs, the overall deviations of dynamic MLC-based IMRT calculated from the deviations of the four weeks were found as 0.55% ± 0.57% and 1.48% ± 0.57% for prostate and H&N patients, respectively. The prostate IMRT plans based on static MLC method had the overall deviations of 1.23% ± 0.69%, 3.07% ± 1.07%, and 3.13% ± 1.29% for intensity levels of IL5, IL10, and IL20, respectively. Moreover, the overall deviations for H&N patients were found as 1.87% ± 0.86%, 3.11% ± 1.24%, and 2.78% ± 1.31% for the static MLC-based IMRT plans with intensity levels of IL5, IL10 and IL20, respectively. Similar with the DAVID results, the error rates in DynaLog files showed upward movement comparing the dynamic IMRT with static IMRT with high intensity levels. In respect to positioning errors higher than 0.005 cm, static prostate IMRT plans with intensity levels of IL10 and IL20 had 1.5 and 2.6 times higher error ratios than dynamic prostate IMRT plans, respectively, while these values stepped up to 8.4 and 12.0 for H&N cases. On the other hand, according to the leaf pair readings, reconstructed dose values from DynaLog files had significant correlation (r = 0.80) with DAVID and EPID readings while a stronger relationship (r = 0.98) was found between the two dosimetric systems. The correlation coefficients for deviations from reference plan readings were found in the interval of -0.21-0.16 for all three systems. The dynamic MLC method showed higher performance in repeatability of leaf positioning than static MLC methods with higher intensity levels even though the deviations in the MLC leaf positioning were found to be under the acceptance threshold for all MLC methods. The high intensity levels increased the position-ing deviations along with the delivery complexity of the static MLC-based IMRT plans. Moreover, DAVID and EPID readings and DynaLog recordings showed mutually strong correlation, while no significant relationship was found between deviations from reference values.

Evaluation of Photoneutron Dose Measured by Bubble Detectors in Conventional Linacs and Cyberknife Unit: Effective Dose and Secondary Malignancy Risk Estimation.

This study aims to reduce the uncertainty about the photoneutron dose produced over a course of radiotherapy with high-energy photon beams and evaluate photoneutron contamination-based secondary malignancy risk for different treatment modalities. Dosimetric measurements were taken in Philips SL25/75, Elekta Synergy Platform (Elekta AB, Stockholm, Sweden), Varian Clinac DHX High Performance systems (Varian Medical Systems, Palo Alto, CA), and Cyberknife Robotic Radiosurgery Unit (Accuray Inc., Sunnyvale, CA) using bubble detector for neutron dosimetry. The measurement data were used to determine in-field and out-of-field neutron equivalent dose in 6-MV 3D conformal radiotherapy, sliding window-intensity-modulated radiotherapy, and stereotactic body radiotherapy and to calculate the effective dose in 18-MV 3D conformal radiotherapy and sliding window-intensity-modulated radiotherapy techniques for patients with prostate cancer undergoing a standard treatment. For the 18-MV treatment techniques, the secondary malignancy risk due to the neutron contamination was estimated using the risk factors published by The International Commission on Radiological Protection. The neutron contamination-based secondary malignancy risk for the 18-MV 3D conformal radiotherapy and sliding window-intensity-modulated radiotherapy modalities was found to be 0.44% and 1.45% for Elekta Synergy Platform and 0.92% and 3.0% for the Varian Clinac DHX High Performance, respectively. For 6-MV 3D conformal radiotherapy, sliding window-intensity-modulated radiotherapy, and stereotactic body radiotherapy treatment techniques, neutron equivalent doses inside the treatment field were found to be lower than 40 mSv. Our measurements reveal that equivalent dose and effective dose due to the neutron contamination are at a considerable level for 18-MV sliding window-intensity-modulated radiotherapy treatments, while 6-MV photon beams used in different modalities still induce only negligible photoneutrons. The secondary malignancy risk based on photoneutron should be therefore taken into consideration in case of selecting 18-MV photons in a sliding window-intensity-modulated radiotherapy treatment instead of 6 MV.

Investigating in-field and out-of-field neutron contamination in high-energy medical linear accelerators based on the treatment factors of field size, depth, beam modifiers, and beam type.

PURPOSE: We analysed the effects of field size, depth, beam modifier and beam type on the amount of in-field and out-of-field neutron contamination for medical linear accelerators (linacs). METHODS: Measurements were carried out for three high-energy medical linacs of Elekta Synergy Platform, Varian Clinac DHX High Performance and Philips SL25 using bubble detectors. The photo-neutron measurements were taken in the first two linacs with 18 MV nominal energy, whereas the electro-neutrons were measured in the three linacs with 9 MeV, 10 MeV, 15 MeV and 18 MeV. RESULTS: The central neutron doses increased with larger field sizes as a dramatic drop off was observed in peripheral areas. Comparing with the jaws-shaped open-field of 10 × 10 cm, the motorised and physical wedges contributed to neutron contamination at central axis by 60% and 18%, respectively. The similar dose increment was observed in MLC-shaped fields. The contributions of MLCs were in the range of 55-59% and 19-22% in Elekta and Varian linacs comparing with 10 × 10 and 20 × 20 cm open fields shaped by the jaws, respectively. The neutron doses at shallow depths were found to be higher than the doses found at deeper regions. The electro-neutron dose at the 18 MeV energy was higher than the doses at the electron energies of 15 MeV and 9 MeV by a factor of 3 and 50, respectively. CONCLUSION: The photo- and electro-neutron dose should be taken into consideration in the radiation treatment with high photon and electron energies.

Verification of dose distribution for volumetric modulated arc therapy total marrow irradiation in a humanlike phantom.

PURPOSE: Volumetric modulated arc therapy (VMAT) treatment planning studies have been reported to provide good target coverage and organs at risk (OARs) sparing in total marrow irradiation (TMI). A comprehensive dosimetric study simulating the clinical situation as close as possible is a norm in radiotherapy before a technique can be used to treat a patient. Without such a study, it would be difficult to make a reliable and safe clinical transition especially with a technique as complicated as VMAT-TMI. To this end, the dosimetric feasibility of VMAT-TMI technique in terms of treatment planning, delivery efficiency, and the most importantly three dimensional dose distribution accuracy was investigated in this study. The VMAT-TMI dose distribution inside a humanlike Rando phantom was measured and compared to the dose calculated using RapidArc especially in the field junctions and the inhomogeneous tissues including the lungs, which is the dose-limiting organ in TMI. METHODS: Three subplans with a total of nine arcs were used to treat the planning target volume (PTV), which was determined as all the bones plus the 3 mm margin. Thermoluminescent detectors (TLDs) were placed at 39 positions throughout the phantom. The measured TLD doses were compared to the calculated plan doses. Planar dose for each arc was verified using mapcheck. RESULTS: TLD readings demonstrated accurate dose delivery, with a median dose difference of 0.5% (range: -4.3% and 6.6%) from the calculated dose in the junctions and in the inhomogeneous medium including the lungs. CONCLUSIONS: The results from this study suggest that RapidArc VMAT technique is dosimetrically accurate, safe, and efficient in delivering TMI within clinically acceptable time frame.

Total marrow irradiation with RapidArc volumetric arc therapy.

PURPOSE: To develop a volumetric arc therapy (VMAT)-total marrow irradiation (TMI) technique for patients with hematologic malignancies. METHODS AND MATERIALS: VMAT planning was performed for 6 patients using RapidArc technology. The planning target volume consisted of all the bones in the body from the head to the mid-femur, excluding the extremities, except for the humerus, plus a 3.0-mm margin. The organs at risk included the lungs, heart, liver, kidneys, bowels, brain, eyes, and oral cavity. The VMAT-TMI technique consisted of three plans: the head and neck, the chest, and the pelvis, each with three 330° arcs. The plans were prescribed to ensure, at a minimum, 95% planning target volume dose coverage with the prescription dose (percentage of volume receiving dose of ≥12 Gy was 95%). The treatments were delivered and verified using MapCheck and ion chamber measurements. RESULTS: The VMAT-TMI technique reported in the present study provided comparable dose distributions with respect to the fixed gantry linear accelerator intensity-modulated TMI. RapidArc planning was less subjective and easier, and, most importantly, the delivery was more efficient. RapidArc reduced the treatment delivery time to approximately 18 min from 45 min with the fixed gantry linear accelerator intensity-modulated TMI. When the prescription dose coverage was reduced to 85% from 95% and the mandible and maxillary structures were not included in the planning target volume as reported in a tomotherapy study, a considerable organ at risk dose reduction of 4.2-51% was observed. The average median dose for the lungs and lenses was reduced to 5.6 Gy from 7.2 Gy and 2.4 Gy from 4.5 Gy, respectively. CONCLUSION: The RapidArc VMAT technique improved the treatment planning, dose conformality, and, most importantly, treatment delivery efficiency. The results from our study suggest that the RapidArc VMAT technology can be expected to facilitate the clinical transition of TMI.

Linear accelerator-based intensity-modulated total marrow irradiation technique for treatment of hematologic malignancies: a dosimetric feasibility study.

PURPOSE: To investigate the dosimetric feasibility of linear accelerator-based intensity-modulated total marrow irradiation (IM-TMI) in patients with hematologic malignancies. METHODS AND MATERIALS: Linear accelerator-based IM-TMI treatment planning was performed for 9 patients using the Eclipse treatment planning system. The planning target volume (PTV) consisted of all the bones in the body from the head to the mid-femur, except for the forearms and hands. Organs at risk (OAR) to be spared included the lungs, heart, liver, kidneys, brain, eyes, oral cavity, and bowel and were contoured by a physician on the axial computed tomography images. The three-isocenter technique previously developed by our group was used for treatment planning. We developed and used a common dose-volume objective method to reduce the planning time and planner subjectivity in the treatment planning process. RESULTS: A 95% PTV coverage with the 99% of the prescribed dose of 12 Gy was achieved for all nine patients. The average dose reduction in OAR ranged from 19% for the lungs to 68% for the lenses. The common dose-volume objective method decreased the planning time by an average of 35% and reduced the inter- and intra- planner subjectivity. CONCLUSION: The results from the present study suggest that the linear accelerator-based IM-TMI technique is clinically feasible. We have demonstrated that linear accelerator-based IM-TMI plans with good PTV coverage and improved OAR sparing can be obtained within a clinically reasonable time using the common dose-volume objective method proposed in the present study.

Normal tissue complication probability modeling of acute hematologic toxicity in cervical cancer patients treated with chemoradiotherapy.

PURPOSE: To test the hypothesis that increased pelvic bone marrow (BM) irradiation is associated with increased hematologic toxicity (HT) in cervical cancer patients undergoing chemoradiotherapy and to develop a normal tissue complication probability (NTCP) model for HT. METHODS AND MATERIALS: We tested associations between hematologic nadirs during chemoradiotherapy and the volume of BM receiving≥10 and 20 Gy (V10 and V20) using a previously developed linear regression model. The validation cohort consisted of 44 cervical cancer patients treated with concurrent cisplatin and pelvic radiotherapy. Subsequently, these data were pooled with data from 37 identically treated patients from a previous study, forming a cohort of 81 patients for normal tissue complication probability analysis. Generalized linear modeling was used to test associations between hematologic nadirs and dosimetric parameters, adjusting for body mass index. Receiver operating characteristic curves were used to derive optimal dosimetric planning constraints. RESULTS: In the validation cohort, significant negative correlations were observed between white blood cell count nadir and V10 (regression coefficient (ß)=-0.060, p=0.009) and V20 (ß=-0.044, p=0.010). In the combined cohort, the (adjusted) ß estimates for log (white blood cell) vs. V10 and V20 were as follows: -0.022 (p=0.025) and -0.021 (p=0.002), respectively. Patients with V10≥95% were more likely to experience Grade≥3 leukopenia (68.8% vs. 24.6%, p<0.001) than were patients with V20>76% (57.7% vs. 21.8%, p=0.001). CONCLUSIONS: These findings support the hypothesis that HT increases with increasing pelvic BM volume irradiated. Efforts to maintain V10<95% and V20<76% may reduce HT.

Probing the existence of medium pulmonary crackles via model-based clustering.

The objective of this study is to probe the existence of a third crackle type, medium, besides the traditionally accepted types, namely, fine and coarse crackles and, furthermore, to explore the representative parameter values for each crackle type. A set of clustering experiments have been conducted on pulmonary crackles to this end. A model-based clustering algorithm, the Expectation-Maximization algorithm, is used and the resulting cluster numbers are validated with Bayesian Inference Criterion. Four different feature sets are extracted from the preprocessed crackle samples, the first of which consists of conventional parameters derived from the zero-crossings of crackle waveforms. The second feature set corresponds to the spectral components of the crackles, whereas the remaining two sets are derived from a single- and double-nodes wavelet network modeling. The results of the clustering experiments demonstrate strong evidence for the existence of a third crackle type. Moreover the labels yielded by clustering experiments, using different feature sets match for roughly 80% or more of the crackle samples, resulting in similar representative crackle parameter values of the three clusters for all feature sets.

Feature extraction for pulmonary crackle representation via wavelet networks.

In this study, wavelet networks have been used to parameterize and quantify pulmonary crackles with an aim to depict the waveform with a small set of meaningful parameters. Complex Morlet wavelets are used at the nodes of both single and double-node networks to model the waveforms with the double-node rendering smaller modeling error. The features extracted from the model parameters have been compared with the conventional time domain features in a two-class clustering experiment with nearly 90% matching between the clusters of different parameter sets and with the model parameters forming clusters more closely distributed around their means and better separated from each other. Moreover, using simulated crackles embedded on real respiratory sounds, features extracted from wavelet networks have been shown to be more robust to background vesicular sounds compared to conventional parameters which are very sensitive to noise.

Elimination of vesicular sounds from pulmonary crackle waveforms.

Pulmonary crackles and their parameters are very useful in the diagnosis of pulmonary disorders. A new automatic method has been proposed for the elimination of background vesicular sound from crackle signal with a view to introduce minimum distortion to crackle parameters. A region of interest is designated and a distortion metric based on the correlation between raw and filtered waveforms in that region is defined. Filter cut-off frequency is estimated based on the distortion metric. To reduce computational cost, a regression analysis is also realized which predicts a new fitted cut-off frequency from the estimated cut-off frequency. As a comparison basis, wavelet filtering is also applied on the same data. The algorithm is validated on simulated crackles superimposed on recorded vesicular sound with results indicating that filtering is achieved with minimal distortion of crackle parameters. The algorithm is also applied on real crackles from subjects with various respiratory disorders. The results show the extent of the effect of vesicular sound on crackle parameters, emphasizing the significance of proper filtering in crackle studies.

Sensitivity of pulmonary crackle parameters to filter cut-off frequency.

Pulmonary crackles are very important indicators in the diagnosis of pulmonary disorders. Parameters derived from their waveforms, such as initial deflection width and two-cycle duration width, are useful in classifying a crackle and correlating it with various disorders. To obtain a crackle waveform, the background vesicular sound is filtered out and the crackle parameters are very sensitive to filter cut-off frequency. In this study, sensitivity analysis of waveform parameters of simulated crackles superimposed on real vesicular sound is carried out, emphasizing the necessity of precision in determining the filter cut-off frequency.

Classifying respiratory sounds with different feature sets.

In this study, different feature sets are used in conjunction with (k-nearest neighbors) k-NN and artificial neural network (ANN) classifiers to address the classification problem of respiratory sound signals. A comparison is made between the performances of k-NN and ANN classifiers with different feature sets derived from respiratory sound data acquired from one microphone placed on the posterior chest area. Each subject is represented by a single respiration cycle divided into sixty segments from which three different feature sets consisting of 6th order AR model coefficients, wavelet coefficients and crackle parameters in addition to AR model coefficients are extracted. Classification experiments are carried out on inspiration and expiration phases separately. The two class recognition problem between healthy and pathological subjects is addressed.