Deep Bleeder Acoustic Coagulation (DBAC)-part II: in vivo testing of a research prototype system

Background

Deep Bleeder Acoustic Coagulation (DBAC) is an ultrasound image-guided high-intensity focused ultrasound (HIFU) method proposed to automatically detect and localize (D&L) and treat deep, bleeding, combat wounds in the limbs of soldiers. A prototype DBAC system consisting of an applicator and control unit was developed for testing on animals. To enhance control, and thus safety, of the ultimate human DBAC autonomous product system, a thermal coagulation strategy that minimized cavitation, boiling, and non-linear behaviors was used.

Material and methods

The in vivo DBAC applicator design had four therapy tiles (Tx) and two 3D (volume) imaging probes (Ix) and was configured to be compatible with a porcine limb bleeder model developed in this research. The DBAC applicator was evaluated under quantitative test conditions (e.g., bleeder depths, flow rates, treatment time limits, and dose exposure time limits) in an in vivo study (final exam) comprising 12 bleeder treatments in three swine. To quantify blood flow rates, the “bleeder” targets were intact arterial branches, i.e., the superficial femoral artery (SFA) and a deep femoral artery (DFA). D&L identified, characterized, and targeted bleeders. The therapy sequence selected Tx arrays and determined the acoustic power and Tx beam steering, focus, and scan patterns. The user interface commands consisted of two buttons: “Start D&L” and “Start Therapy.” Targeting accuracy was assessed by necropsy and histologic exams and efficacy (vessel coagulative occlusion) by angiography and histology.

Results

The D&L process (Part I article, J Ther Ultrasound, 2015 (this issue)) executed fully in all cases in under 5 min and targeting evaluation showed 11 of 12 thermal lesions centered on the correct vessel subsection, with minimal damage to adjacent structures. The automated therapy sequence also executed properly, with select manual steps. Because the dose exposure time limit (t_dose ≤ 30 s) was associated with nonefficacious treatment, 60-s dosing and dual-dosing was also pursued. Thrombogenic evidence (blood clotting) and collagen denaturation (vessel shrinkage) were found in necropsy and histologically in all targeted SFAs. Acute SFA reductions in blood flow (20–30 %) were achieved in one subject, and one partial and one complete vessel occlusion were confirmed angiographically. The complete occlusion case was achieved with a dual dose (90 s total exposure) with focal intensity ≈500 W/cm² (spatial average, temporal average).

Conclusions

While not meeting all in vivo objectives, the overall performance of the DBAC applicator was positive. In particular, D&L automation workflow was verified during each of the tests, with processing times well under specified (10 min) limits, and all bleeder branches were detected and localized. Further, gross necropsy and tissue examination confirmed that the HIFU thermal lesions were coincident with the target vessel locations in over 90 % of the multi-array dosing treatments. The SFA/DFA bleeder models selected, and the protocols used, were the most suitable practical model options for the given DBAC anatomical and bleeder requirements. The animal models were imperfect in some challenging aspects, including requiring tissue-mimicking material (TMM) standoffs to achieve deep target depths, thereby introducing device-tissue motion, with resultant imaging artifacts. The model “bleeders” involved intact vessels, which are subject to less efficient heating and coagulation cascade behaviors than true puncture injuries.

Deep bleeder acoustic coagulation (DBAC)-Part I: development and in vitro testing of a research prototype cuff system

Background

Bleeding from limb injuries is a leading cause of death on the battlefield, with deep wounds being least accessible. High-intensity focused ultrasound (HIFU) has been shown capable of coagulation of bleeding (cautery). This paper describes the development and refereed in vitro evaluation of an ultrasound (US) research prototype deep bleeder acoustic coagulation (DBAC) cuff system for evaluating the potential of DBAC in the battlefield. The device had to meet quantitative performance metrics on automated operation, therapeutic heating, bleeder detection, targeting accuracy, operational time limits, and cuff weight over a range of limb sizes and bleeder depths. These metrics drove innovative approaches in image segmentation, bleeder detection, therapy transducers, beam targeting, and dose monitoring. A companion (Part II) paper discusses the in vivo performance testing of an animal-specific DBAC system.

Materials and methods

The cuff system employed 3D US imaging probes (“Ix”) for detection and localization (D&L) and targeting, with the bleeders being identified by automated spectral Doppler analysis of flow waveforms. Unique high-element-count therapeutic arrays (“Tx”) were developed, with the final cuff prototype having 21 Tx’s and 6 Ix’s. Spatial registration of Ix’s and Tx’s was done with a combination of image-registration, acoustic time-of-flight measurement, and tracking of the cuff shape via a fiber optic sensor. Acoustic radiation force impulse (ARFI) imaging or thermal strain imaging (TSI) at low-power doses were used to track the HIFU foci in closed-loop targeting. Recurrent neural network (RNN) acoustic thermometry guided closed-loop dosing. The cuff was tested on three phantom “limb” sizes: diameters = 25, 15, and 7.5 cm, with bleeder depths from 3.75 to 12.5 cm. “Integrated Phantoms” (IntP) were used for assessing D&L, closed-loop targeting, and closed-loop dosing. IntPs had surrogate arteries and bleeders, with blood-mimicking fluids moved by a pulsatile pump, and thermocouples (TCs) on the bleeders. Acoustic dosing was developed and tested using “HIFU Phantoms” having precisely located TCs, with end-of-dose target ∆T = 33–58 °C, and skin temperature ∆T ≤ 20 °C, being required.

Results

Most DBAC cuff performance requirements were met, including cuff weight, power delivery, targeting accuracy, skin temperature limit, and autonomous operation. The automated D&L completed in 9 of 15 tests (65 %), detecting the smallest (0.6 mm) bleeders, but it had difficulty with the lowest flow (3 cm/sec) bleeders, and in localizing bleeders in the smallest (7.5 cm) phantoms. D&L did not complete within the 9-min limit (results ranged 10–21 min). Closed-loop targeting converged in 20 of 31 tests (71 %), and closed-loop dosing power shut-off at preset ∆Ts was operational.

Summary and conclusion

The main performance objectives of the prototype DBAC cuff were met, however the designs required a number of challenging new technology developments. The novel Tx arrays exhibited high power with significant beam steering and focusing flexibility, while their integrated electronics enabled the required compact, lightweight configurability and simplified driving controls and cable/connector architecture. The compounded 3D imaging, combined with sophisticated software algorithms, enabled automated D&L and initial targeting and closed-loop targeting feedback via TSI. The development of RNN acoustic thermometry made possible feedback-controlled dosing. The lightweight architecture required significant design and fabrication effort to meet mechanical functionalities. Although not all target specifications were met, future engineering solutions addressing these performance deficiencies are proposed. Lastly, the program required very complex limb test phantoms and, while very challenging to develop, they performed well.

Robust and fast contrast inflow detection for 2D X-ray fluoroscopy

2D X-ray fluoroscopy is widely used in computer assisted and image guided interventions because of the real time visual guidance it can provide to the physicians. During cardiac interventions, acquisitions of angiography are often used to assist the physician in visualizing the blood vessel structures, guide wires, or catheters, localizing bifurcations, estimating severity of a lesion, or observing the blood flow. Computational algorithms often need to process differently to frames with or without contrast medium. In order to automate this process and streamline the clinical workflow, a fully automatic contrast inflow detection algorithm is proposed. The robustness of the algorithm is validated by more than 1300 real fluoroscopic scenes. The algorithm is computationally efficient; a sequence with 100 frames can be processed within a second.

A review of 3D vessel lumen segmentation techniques: models, features and extraction schemes

Vascular diseases are among the most important public health problems in developed countries. Given the size and complexity of modern angiographic acquisitions, segmentation is a key step toward the accurate visualization, diagnosis and quantification of vascular pathologies. Despite the tremendous amount of past and on-going dedicated research, vascular segmentation remains a challenging task. In this paper, we review state-of-the-art literature on vascular segmentation, with a particular focus on 3D contrast-enhanced imaging modalities (MRA and CTA). We structure our analysis along three axes: models, features and extraction schemes. We first detail model-based assumptions on the vessel appearance and geometry which can embedded in a segmentation approach. We then review the image features that can be extracted to evaluate these models. Finally, we discuss how existing extraction schemes combine model and feature information to perform the segmentation task. Each component (model, feature and extraction scheme) plays a crucial role toward the efficient, robust and accurate segmentation of vessels of interest. Along each axis of study, we discuss the theoretical and practical properties of recent approaches and highlight the most advanced and promising ones.

Liver segmentation using sparse 3D prior models with optimal data support

Volume segmentation is a relatively slow process and, in certain circumstances, the enormous amount of prior knowledge available is underused. Model-based liver segmentation suffers from the large shape variability of this organ, and from structures of similar appearance that juxtapose the liver. The technique presented in this paper is devoted to combine a statistical analysis of the data with a reconstruction model from sparse information: only the most reliable information in the image is used, and the rest of the liver's shape is inferred from the model and the sparse observation. The resulting process is more efficient than standard segmentation since most of the workload is concentrated on the critical points, but also more robust, since the interpolated volume is consistent with the prior knowledge statistics. The experimental results on liver datasets prove the sparse information model has the same potential as PCA, if not better, to represent the shape of the liver. Furthermore, the performance assessment from measurement statistics on the liver's volume, distance between reconstructed surfaces and ground truth, and inter-observer variability demonstrates the liver is efficiently segmented using sparse information.

An energy minimization approach to the data driven editing of presegmented images/volumes

Fully automatic, completely reliable segmentation in medical images is an unrealistic expectation with today's technology. However, many automatic segmentation algorithms may achieve a near-correct solution, incorrect only in a small region. For these situations, an interactive editing tool is required, ideally in 3D, that is usually left to a manual correction. We formulate the editing task as an energy minimization problem that may be solved with a modified version of either graph cuts or the random walker 3D segmentation algorithms. Both algorithms employ a seeded user interface, that may be used in this scenario for a user to seed erroneous voxels as belonging to the foreground or the background. In our formulation, it is unnecessary for the user to specify both foreground and background seeds.

4D shape priors for a level set segmentation of the left myocardium in SPECT sequences

We develop a 4D (3D plus time) statistical shape model for implicit level set based shape representations. To this end, we represent hand segmented training sequences of the left ventricle by respective 4-dimensional embedding functions and approximate these by a principal component analysis. In contrast to recent 4D models on explicit shape representations, the implicit shape model developed in this work does not require the computation of point correspondences which is known to be quite challenging, especially in higher dimensions. Experimental results on the segmentation of SPECT sequences of the left myocardium confirm that the 4D shape model outperforms respective 3D models, because it takes into account a statistical model of the temporal shape evolution.

Soft level set coupling for LV segmentation in gated perfusion SPECT

We present a new segmentation approach for the myocardium in gated and non-gated perfusion SPECT images. To this end, we represent the epi- and endocardium by separate signed distance functions and couple them by a soft constraint to give explicit control over the wall thickness. By an explicit modeling of the basal plane, the volume of the blood pool as well as the myocardium are determinable. Furthermore, prior shape information is incorporated by applying a kernel density estimation on a number of expert segmentations in a low-dimensional PCA subspace. Thereby, information along the time axis is fully taken into account by employing 4-dimensional embedding functions.