Search

Hyperthermia treatment planning is a deeply patient-specific task which includes the optimal determination of the excitations of an array applicator. To enhance flexibility, various solutions exploiting different frequencies, antenna element, number, and applicator geometries have been proposed in the literature. Amongst them, increasing the frequency and the number of radiating elements have been effective for achieving more conformal heating. However, as each radiating element requires a power amplifier to control it, increasing the number of antennas considerably impacts the overall cost and complexity of the system. Accordingly, a procedure capable of selecting an optimal patient-specific subset of antennas from an oversized phased array applicator (with more antenna elements than available amplifiers) could help improve cost-effectiveness. In this study, we present an original approach, which allows improving performance by adaptively selecting the optimal subset of antennas to be activated for a given (redundant) applicator and a given patient. The proposed approach takes inspiration from the compressive sensing theory by embedding the sparsity promotion paradigm into a treatment planning procedure, which casts power deposition as a constrained convex optimization. Performances were demonstrated for the case of head and neck hyperthermia and benchmarked against the antenna selection procedure presently used in clinical practice.

Purpose: Hyperthermia treatment quality determines treatment effectiveness as shown by the clinically derived thermal-dose effect relations. SAR based optimization factors are used as possible surrogate for temperature, since they are not affected by thermal tissue properties uncertainty and variations. Previously, target coverage (TC) at the 25% and 50% iso-SAR level was shown predictive for treatment outcome in superficial hyperthermia and the target-to-hot-spot-quotient (THQ) was shown to highly correlate with predictive temperature in deep pelvic hyperthermia. Here, we investigate the correlation with temperature for THQ and TC using an ‘intermediate’ scenario: semi-deep hyperthermia in the head & neck region using the HYPERcollar3D. Methods: Fifteen patient-specific models and two different planning approaches were used, including random perturbations to circumvent optimization bias. The predicted SAR indicators were compared to predicted target temperature distribution indicators T50 and T90, i.e., the median and 90th percentile temperature respectively. Results: The intra-patient analysis identified THQ, TC25 and TC50 as good temperature surrogates: with a mean correlation coefficient R 2 T50 = 0.72 and R 2 T90 =0.66. The inter-patient analysis identified the highest correlation with TC25 (R 2 T50 = 0.76, R 2 T90 =0.54) and TC50 (R 2 T50 = 0.74, R 2 T 90 = 0.56). Conclusion: Our investigation confirmed the validity of our current strategy for deep hyperthermia in the head & neck based on a combination of THQ and TC25. TC50 was identified as the best surrogate since it enables optimization and patient inclusion decision making using one single parameter.

Hyperthermia treatment planning is a deeply patient-specific task which includes the optimal determination of the excitations of an array applicator. To enhance flexibility, various solutions exploiting different frequencies, antenna element, number, and applicator geometries have been proposed in the literature. Amongst them, increasing the frequency and the number of radiating elements have been effective for achieving more conformal heating. However, as each radiating element requires a power amplifier to control it, increasing the number of antennas considerably impacts the overall cost and complexity of the system. Accordingly, a procedure capable of selecting an optimal patient-specific subset of antennas from an oversized phased array applicator (with more antenna elements than available amplifiers) could help improve cost-effectiveness. In this study, we present an original approach, which allows improving performance by adaptively selecting the optimal subset of antennas to be activated for a given (redundant) applicator and a given patient. The proposed approach takes inspiration from the compressive sensing theory by embedding the sparsity promotion paradigm into a treatment planning procedure, which casts power deposition as a constrained convex optimization. Performances were demonstrated for the case of head and neck hyperthermia and benchmarked against the antenna selection procedure presently used in clinical practice.

Purpose: Hyperthermia treatment quality determines treatment effectiveness as shown by the clinically derived thermal-dose effect relations. SAR based optimization factors are used as possible surrogate for temperature, since they are not affected by thermal tissue properties uncertainty and variations. Previously, target coverage (TC) at the 25% and 50% iso-SAR level was shown predictive for treatment outcome in superficial hyperthermia and the target-to-hot-spot-quotient (THQ) was shown to highly correlate with predictive temperature in deep pelvic hyperthermia. Here, we investigate the correlation with temperature for THQ and TC using an ‘intermediate’ scenario: semi-deep hyperthermia in the head & neck region using the HYPERcollar3D. Methods: Fifteen patient-specific models and two different planning approaches were used, including random perturbations to circumvent optimization bias. The predicted SAR indicators were compared to predicted target temperature distribution indicators T50 and T90, i.e., the median and 90th percentile temperature respectively. Results: The intra-patient analysis identified THQ, TC25 and TC50 as good temperature surrogates: with a mean correlation coefficient R 2 T50 = 0.72 and R 2 T90 =0.66. The inter-patient analysis identified the highest correlation with TC25 (R 2 T50 = 0.76, R 2 T90 =0.54) and TC50 (R 2 T50 = 0.74, R 2 T 90 = 0.56). Conclusion: Our investigation confirmed the validity of our current strategy for deep hyperthermia in the head & neck based on a combination of THQ and TC25. TC50 was identified as the best surrogate since it enables optimization and patient inclusion decision making using one single parameter.

International audience, This communication deals with the reconstruction of three- dimensional target from experimental multiple-frequency data measured in the anechoic chamber of the Institut Fresnel (Marseille, France). In order to take into account the random noise present in the experiments, a Bayesian approach is considered. In particular, the inversion procedure takes advantage from the joint use of the Contrast Source-Extended Born model and of a Markov Random Field regularization. We also considered a cost functional appropriately weighted by coefficients which change with the frequency, the incident angle and the receiving angle. In fact, each scattered field measurement is balanced with the noise disturbing the data.

International audience

The synthesis of sum and difference patterns is a canonical problem widely dealt with by researchers working on antenna array synthesis. As a matter of fact, they are used as transmitting/receiving devices for search‐and‐track systems (e.g., monopulse radars [1]). In this framework, several procedures have been proposed to reduce the complexity of the beam forming network aimed at generating at least a couple of radiation patterns. Among them, the generation of an optimal sum pattern and a difference one has been carried out by means of a sub‐arraying strategy [2,3]. The simplification of the hardware complexity has also been addressed by sharing some excitations for the sum and difference channels [4]. Recently, the synthesis of low‐sidelobe sum and difference patterns with a common aperture has been carried out by perturbing the roots of the Bayliss distribution to match as much as possible a given Taylor distribution [5]. The discrete linear arrays have been successively obtained by sampling the resulting continuous apertures. In this work, the same array synthesis problem dealt with in [5] is addressed, and an innovative approach based on a deterministic optimization strategy is presented wherein the problem is formulated as the minimization of a linear function over a convex set. Taking advantage from the approaches proposed in [6] and [7] for the optimal synthesis of sum and difference patterns respectively, the proposed method allows one to synthesize patterns with arbitrary sidelobes (unlike [5]).

The synthesis of sum and difference patterns is a canonical problem widely dealt with by researchers working on antenna array synthesis. As a matter of fact, they are used as transmitting/receiving devices for search‐and‐track systems (e.g., monopulse radars [1]). In this framework, several procedures have been proposed to reduce the complexity of the beam forming network aimed at generating at least a couple of radiation patterns. Among them, the generation of an optimal sum pattern and a difference one has been carried out by means of a sub‐arraying strategy [2,3]. The simplification of the hardware complexity has also been addressed by sharing some excitations for the sum and difference channels [4]. Recently, the synthesis of low‐sidelobe sum and difference patterns with a common aperture has been carried out by perturbing the roots of the Bayliss distribution to match as much as possible a given Taylor distribution [5]. The discrete linear arrays have been successively obtained by sampling the resulting continuous apertures. In this work, the same array synthesis problem dealt with in [5] is addressed, and an innovative approach based on a deterministic optimization strategy is presented wherein the problem is formulated as the minimization of a linear function over a convex set. Taking advantage from the approaches proposed in [6] and [7] for the optimal synthesis of sum and difference patterns respectively, the proposed method allows one to synthesize patterns with arbitrary sidelobes (unlike [5]).

The synthesis of sum and difference patterns is a canonical problem widely dealt with by researchers working on antenna array synthesis. As a matter of fact, they are used as transmitting/receiving devices for search‐and‐track systems (e.g., monopulse radars [1]). In this framework, several procedures have been proposed to reduce the complexity of the beam forming network aimed at generating at least a couple of radiation patterns. Among them, the generation of an optimal sum pattern and a difference one has been carried out by means of a sub‐arraying strategy [2,3]. The simplification of the hardware complexity has also been addressed by sharing some excitations for the sum and difference channels [4]. Recently, the synthesis of low‐sidelobe sum and difference patterns with a common aperture has been carried out by perturbing the roots of the Bayliss distribution to match as much as possible a given Taylor distribution [5]. The discrete linear arrays have been successively obtained by sampling the resulting continuous apertures. In this work, the same array synthesis problem dealt with in [5] is addressed, and an innovative approach based on a deterministic optimization strategy is presented wherein the problem is formulated as the minimization of a linear function over a convex set. Taking advantage from the approaches proposed in [6] and [7] for the optimal synthesis of sum and difference patterns respectively, the proposed method allows one to synthesize patterns with arbitrary sidelobes (unlike [5]).

The synthesis of sum and difference patterns is a canonical problem widely dealt with by researchers working on antenna array synthesis. As a matter of fact, they are used as transmitting/receiving devices for search‐and‐track systems (e.g., monopulse radars [1]). In this framework, several procedures have been proposed to reduce the complexity of the beam forming network aimed at generating at least a couple of radiation patterns. Among them, the generation of an optimal sum pattern and a difference one has been carried out by means of a sub‐arraying strategy [2,3]. The simplification of the hardware complexity has also been addressed by sharing some excitations for the sum and difference channels [4]. Recently, the synthesis of low‐sidelobe sum and difference patterns with a common aperture has been carried out by perturbing the roots of the Bayliss distribution to match as much as possible a given Taylor distribution [5]. The discrete linear arrays have been successively obtained by sampling the resulting continuous apertures. In this work, the same array synthesis problem dealt with in [5] is addressed, and an innovative approach based on a deterministic optimization strategy is presented wherein the problem is formulated as the minimization of a linear function over a convex set. Taking advantage from the approaches proposed in [6] and [7] for the optimal synthesis of sum and difference patterns respectively, the proposed method allows one to synthesize patterns with arbitrary sidelobes (unlike [5]).

The synthesis of sum and difference patterns is a canonical problem widely dealt with by researchers working on antenna array synthesis. As a matter of fact, they are used as transmitting/receiving devices for search‐and‐track systems (e.g., monopulse radars [1]). In this framework, several procedures have been proposed to reduce the complexity of the beam forming network aimed at generating at least a couple of radiation patterns. Among them, the generation of an optimal sum pattern and a difference one has been carried out by means of a sub‐arraying strategy [2,3]. The simplification of the hardware complexity has also been addressed by sharing some excitations for the sum and difference channels [4]. Recently, the synthesis of low‐sidelobe sum and difference patterns with a common aperture has been carried out by perturbing the roots of the Bayliss distribution to match as much as possible a given Taylor distribution [5]. The discrete linear arrays have been successively obtained by sampling the resulting continuous apertures. In this work, the same array synthesis problem dealt with in [5] is addressed, and an innovative approach based on a deterministic optimization strategy is presented wherein the problem is formulated as the minimization of a linear function over a convex set. Taking advantage from the approaches proposed in [6] and [7] for the optimal synthesis of sum and difference patterns respectively, the proposed method allows one to synthesize patterns with arbitrary sidelobes (unlike [5]).

The synthesis of sum and difference patterns is a canonical problem widely dealt with by researchers working on antenna array synthesis. As a matter of fact, they are used as transmitting/receiving devices for search‐and‐track systems (e.g., monopulse radars [1]). In this framework, several procedures have been proposed to reduce the complexity of the beam forming network aimed at generating at least a couple of radiation patterns. Among them, the generation of an optimal sum pattern and a difference one has been carried out by means of a sub‐arraying strategy [2,3]. The simplification of the hardware complexity has also been addressed by sharing some excitations for the sum and difference channels [4]. Recently, the synthesis of low‐sidelobe sum and difference patterns with a common aperture has been carried out by perturbing the roots of the Bayliss distribution to match as much as possible a given Taylor distribution [5]. The discrete linear arrays have been successively obtained by sampling the resulting continuous apertures. In this work, the same array synthesis problem dealt with in [5] is addressed, and an innovative approach based on a deterministic optimization strategy is presented wherein the problem is formulated as the minimization of a linear function over a convex set. Taking advantage from the approaches proposed in [6] and [7] for the optimal synthesis of sum and difference patterns respectively, the proposed method allows one to synthesize patterns with arbitrary sidelobes (unlike [5]).

The synthesis of sum and difference patterns is a canonical problem widely dealt with by researchers working on antenna array synthesis. As a matter of fact, they are used as transmitting/receiving devices for search‐and‐track systems (e.g., monopulse radars [1]). In this framework, several procedures have been proposed to reduce the complexity of the beam forming network aimed at generating at least a couple of radiation patterns. Among them, the generation of an optimal sum pattern and a difference one has been carried out by means of a sub‐arraying strategy [2,3]. The simplification of the hardware complexity has also been addressed by sharing some excitations for the sum and difference channels [4]. Recently, the synthesis of low‐sidelobe sum and difference patterns with a common aperture has been carried out by perturbing the roots of the Bayliss distribution to match as much as possible a given Taylor distribution [5]. The discrete linear arrays have been successively obtained by sampling the resulting continuous apertures. In this work, the same array synthesis problem dealt with in [5] is addressed, and an innovative approach based on a deterministic optimization strategy is presented wherein the problem is formulated as the minimization of a linear function over a convex set. Taking advantage from the approaches proposed in [6] and [7] for the optimal synthesis of sum and difference patterns respectively, the proposed method allows one to synthesize patterns with arbitrary sidelobes (unlike [5]).