Robust surface registration using salient anatomical features for image-guided liver surgery: algorithm and validation

A successful surface-based image-to-physical space registration in image-guided liver surgery (IGLS) is critical to provide reliable guidance information to surgeons and pertinent surface displacement data for use in deformation correction algorithms. The current protocol used to perform the image-to-physical space registration involves an initial pose estimation provided by a point based registration of anatomical landmarks identifiable in both the preoperative tomograms and the intraoperative presentation. The surface based registration is then performed via a traditional iterative closest point (ICP) algorithm between the preoperative liver surface, segmented from the tomographic image set, and an intraoperatively acquired point cloud of the liver surface provided by a laser range scanner. Using this more conventional method, the registration accuracy can be compromised by poor initial pose estimation as well as tissue deformation due to the laparotomy and liver mobilization performed prior to tumor resection. In order to increase the robustness of the current surface-based registration method used in IGLS, we propose the incorporation of salient anatomical features, identifiable in both the preoperative image sets and intraoperative liver surface data, to aid in the initial pose estimation and play a more significant role in the surface-based registration via a novel weighting scheme. Examples of such salient anatomical features are the falciform groove region as well as the inferior ridge of the liver surface. In order to validate the proposed weighted patch registration method, the alignment results provided by the proposed algorithm using both single and multiple patch regions were compared with the traditional ICP method using six clinical datasets. Robustness studies were also performed using both phantom and clinical data to compare the resulting registrations provided by the proposed algorithm and the traditional method under conditions of varying initial pose. The results provided by the robustness trials and clinical registration comparisons suggest that the proposed weighted patch registration algorithm provides a more robust method with which to perform the image-to-physical space registration in IGLS. Furthermore, the implementation of the proposed algorithm during surgical procedures does not impose significant increases in computation or data acquisition times.     

Non-rigid registration between 3D ultrasound and CT images of the liver based on intensity and gradient information

In order to utilize both ultrasound (US) and computed tomography (CT) images of the liver concurrently for medical applications such as diagnosis and image-guided intervention, non-rigid registration between these two types of images is an essential step, as local deformation between US and CT images exists due to the different respiratory phases involved and due to the probe pressure that occurs in US imaging. This paper introduces a voxel-based non-rigid registration algorithm between the 3D B-mode US and CT images of the liver. In the proposed algorithm, to improve the registration accuracy, we utilize the surface information of the liver and gallbladder in addition to the information of the vessels inside the liver. For an effective correlation between US and CT images, we treat those anatomical regions separately according to their characteristics in US and CT images. Based on a novel objective function using a 3D joint histogram of the intensity and gradient information, vessel-based non-rigid registration is followed by surface-based non-rigid registration in sequence, which improves the registration accuracy. The proposed algorithm is tested for ten clinical datasets and quantitative evaluations are conducted. Experimental results show that the registration error between anatomical features of US and CT images is less than 2 mm on average, even with local deformation due to different respiratory phases and probe pressure. In addition, the lesion registration error is less than 3 mm on average with a maximum of 4.5 mm that is considered acceptable for clinical applications.     

CT-MRI automatic surface-based registration schemes combining global and local optimization techniques.

Medical image registration is commonly required in order to combine the complementary information provided by different medical imaging modalities. In this paper, a new automatic registration scheme is proposed to register 3-D CT-MR head images and is currently tested on a clinical environment. The proposed scheme, after the preprocessing and the outer surface extraction of the data, is based on the use the rigid transformation method, in conjunction with a combination of global and local optimization techniques. Analytically, the paper exploits the optimization efficiency of three widely used optimization techniques, in obtaining the parameters of the rigid transformation model: the Downhill Simplex Method, the Genetic Algorithms and the Simulated Annealing. These optimization techniques are further combined by the sequential application of the Powell optimization method in order to refine the registration and increase its accuracy. A comparative study involving these optimization techniques in conjunction with the rigid transformation, and two other methods, the ICP and the manual methods, is also presented, for a sufficient number of clinical CT-MR brain images. Finally, quantitative and qualitative results are also presented to validate the performance of these automatic surface-based registration schemes, in terms of consistency and accuracy. Throughout of this study, the automatic registration scheme comprising of the rigid transformation in conjunction with the Simulated Annealing method sequentially combined with the Powell method has been performed superior regarding all the other compared registration schemes.     

Rigid 2D/3D slice-to-volume registration and its application on fluoroscopic CT images

Registration of single slices from FluoroCT, CineMR, or interventional magnetic resonance imaging to three dimensional (3D) volumes is a special aspect of the two-dimensional (2D)/3D registration problem. Rather than digitally rendered radiographs (DRR), single 2D slice images obtained during interventional procedures are compared to oblique reformatted slices from a high resolution 3D scan. Due to the lack of perspective information and the different imaging geometry, convergence behavior differs significantly from 2D/3D registration applications comparing DRR images with conventional x-ray images. We have implemented a number of merit functions and local and global optimization algorithms for slice-to-volume registration of computed tomography (CT) and FluoroCT images. These methods were tested on phantom images derived from clinical scans for liver biopsies. Our results indicate that good registration accuracy in the range of 0.50 and 1.0 mm is achievable using simple cross correlation and repeated application of local optimization algorithms. Typically, a registration took approximately 1 min on a standard personal computer. Other merit functions such as pattern intensity or normalized mutual information did not perform as well as cross correlation in this initial evaluation. Furthermore, it appears as if the use of global optimization algorithms such as simulated annealing does not improve reliability or accuracy of the registration process. These findings were also confirmed in a preliminary registration study on five clinical scans. These experiments have, however, shown that a strict breath-hold protocol is inevitable when using rigid registration techniques for lesion localization in image-guided biopsy retrieval. Finally, further possible applications of slice-to-volume registration are discussed.     

Automatic lumbar vertebral identification using surface-based registration

This work proposes the use of surface-based registration to automatically select a particular vertebra of interest during surgery. Manual selection of the correct vertebra can be a challenging task, especially for closed-back, minimally invasive procedures. Our method uses shape variations that exist among lumbar vertebrae to automatically determine the portion of the spinal column surface that correctly matches a set of physical vertebral points. In our experiments, we register vertebral points representing posterior elements of a single vertebra in physical space to spinal column surfaces extracted from computed tomography images of multiple vertebrae. After registering the set of physical points to each vertebral surface that is a potential match, we then compute the standard deviation of the surface error for each registration trial. The registration that corresponds to the lowest standard deviation designates the correct match. We have performed our current experiments on two plastic spine phantoms and two patients.     

Comparative Evaluation of Pelvic Allograft Selection Methods

This paper presents a firsthand comparative evaluation of three different existing methods for selecting a suitable allograft from a bone storage bank. The three examined methods are manual selection, automatic volume-based registration, and automatic surface-based registration. Although the methods were originally published for different bones, they were adapted to be systematically applied on the same data set of hemi-pelvises. A thorough experiment was designed and applied in order to highlight the advantages and disadvantages of each method. The methods were applied on the whole pelvis and on smaller fragments, thus producing a realistic set of clinical scenarios. Clinically relevant criteria are used for the assessment such as surface distances and the quality of the junctions between the donor and the receptor. The obtained results showed that both automatic methods outperform the manual counterpart. Additional advantages of the surface-based method are in the lower computational time requirements and the greater contact surfaces where the donor meets the recipient.     

Automatic Prospective Registration of High-Resolution Trabecular Bone Images of the Tibia

Magnetic Resonance Imaging (MRI) longitudinal studies conducted to assess changes in tibia bone quality impose strict requirements on the reproducibility of the prescribed region acquired. Registration, the process of aligning two images, is commonly performed on the images after acquisition. However, techniques to improve image registration precision by adjusting scanning parameters prospectively, prior to image acquisition, would be preferred. We have adapted an automatic prospective mutual information based registration algorithm to a MRI longitudinal study of trabecular bone of the tibia and compared it to a post-scan manual registration. Qualitatively, image alignment due to the prospective registration is shown in 2D subtraction images and 3D surface renderings. Quantitatively, the registration performance is demonstrated by calculating the sum of the squares of the subtraction images. Results show that the sum of the squares is lower for the follow up images with prospective registration by an average of 19.37% ± 0.07 compared to follow up images with post-scan manual registration. Our study found no significant difference between the trabecular bone structure parameters calculated from the post-scan manual registration and the prospective registration images (p > 0.05). All coefficient of variation values for all trabecular bone structure parameters were within a 2–4.5% range which are within values previously reported in the literature. Results suggest that this algorithm is robust enough to be used in different musculoskeletal imaging applications including the hip as well as the tibia.     

Tracking moving objects with megavoltage portal imaging: a feasibility study

Four different algorithms were investigated with the aim to determine their suitability to track an object in conventional megavoltage portal images. The algorithms considered were the mean of the sum of squared differences (MSSD), mutual information (MI), the correlation ratio (CR), and the correlation coefficient (CC). Simulation studies were carried out with various image series containing a rigid object of interest that was moved along a predefined trajectory. For each of the series the signal-to-noise ratio (SNR) was varied to compare the performance of the algorithms under noisy conditions. For a poor SNR of -6 dB the mean tracking error was 2.4, 6.5, 39.0, and 17.2 pixels for MSSD, CC, CR and MI, respectively, with a standard deviation of 1.9, 12.9, 19.5, and 7.5 pixels, respectively. The size of a pixel was 0.5 mm. These results improved to 1.1, 1.3, 1.3, and 2.0 pixels, respectively, with a standard deviation of 0.6, 0.8, 0.8, and 2.1 pixels, respectively, when a mean filter was applied to the images prior to tracking. The implementation of MSSD into existing in-house software demonstrated that, depending on the search range, it was possible to process between 2 and 15 images/s, making this approach capable of real-time applications. In conclusion, the best geometric tracking accuracy overall was obtained with MSSD, followed by CC, CR, and MI. The simplest and best algorithm, both in terms of geometric accuracy as well as computational cost, was the MSSD algorithm and was therefore the method of choice.     

Efficient nonlinear registration of 3D images using high order co-ordinate transfer functions

There is an increasing interest in image registration for a variety of medical imaging applications. Image registration is achieved through the use of a co-ordinate transfer function (CTF) which maps voxels in one image to voxels in the other image, including in the general case changes in mapped voxel intensity. If images of the same subject are to be registered the coordinate transfer function needs to implement a spatial transformation consisting of a displacement and a rigid rotation. In order to achieve registration a common approach is to choose a suitable quality-of-registration measure and devise a method for the efficient generation of the parameters of the CTF which minimize this measure. For registration of images from different subjects more complex transforms are required. In general function minimization is too slow to allow the use of CTFs with more than a small number of parameters. However, provided the images are from the same modality and the CTF can be expanded in terms of an appropriate set of basis functions this paper will show how relatively complex CTFs can be used for registration. The use of increasingly complex CTFs to minimize the within group standard deviation of a set of normal single photon emission tomography brain images is used to demonstrate the improved registration of images from different subjects using CTFs of increasing complexity.     

A frequency-based approach to locate common structure for 2D-3D intensity-based registration of setup images in prostate radiotherapy

In many radiotherapy clinics, geometric uncertainties in the delivery of 3D conformal radiation therapy and intensity modulated radiation therapy of the prostate are reduced by aligning the patient's bony anatomy in the planning 3D CT to corresponding bony anatomy in 2D portal images acquired before every treatment fraction. In this paper, we seek to determine if there is a frequency band within the portal images and the digitally reconstructed radiographs (DRRs) of the planning CT in which bony anatomy predominates over non-bony anatomy such that portal images and DRRs can be suitably filtered to achieve high registration accuracy in an automated 2D-3D single portal intensity-based registration framework. Two similarity measures, mutual information and the Pearson correlation coefficient were tested on carefully collected gold-standard data consisting of a kilovoltage cone-beam CT (CBCT) and megavoltage portal images in the anterior-posterior (AP) view of an anthropomorphic phantom acquired under clinical conditions at known poses, and on patient data. It was found that filtering the portal images and DRRs during the registration considerably improved registration performance. Without filtering, the registration did not always converge while with filtering it always converged to an accurate solution. For the pose-determination experiments conducted on the anthropomorphic phantom with the correlation coefficient, the mean (and standard deviation) of the absolute errors in recovering each of the six transformation parameters were Theta(x):0.18(0.19) degrees, Theta(y):0.04(0.04) degrees, Theta(z):0.04(0.02) degrees, t(x):0.14(0.15) mm, t(y):0.09(0.05) mm, and t(z):0.49(0.40) mm. The mutual information-based registration with filtered images also resulted in similarly small errors. For the patient data, visual inspection of the superimposed registered images showed that they were correctly aligned in all instances. The results presented in this paper suggest that robust and accurate registration can be achieved with intensity-based methods by focusing on rigid bony structures in the images while diminishing the influence of artifacts with similar frequencies as soft tissue.     

Efficient nonlinear registration of 3D images using high order co-ordinate transfer functions

There is an increasing interest in image registration for a variety of medical imaging applications. Image registration is achieved through the use of a co-ordinate transfer function (CTF) which maps voxels in one image to voxels in the other image, including in the general case changes in mapped voxel intensity. If images of the same subject are to be registered the co-ordinate transfer function needs to implement a spatial transformation consisting of a displacement and a rigid rotation. In order to achieve registration a common approach is to choose a suitable quality-of-registration measure and devise a method for the efficient generation of the parameters of the CTF which minimize this measure. For registration of images from different subjects more complex transforms are required. In general function minimization is too slow to allow the use of CTFs with more than a small number of parameters. However, provided the images are from the same modality and the CTF can be expanded in terms of an appropriate set of basis functions this paper will show how relatively complex CTFs can be used for registration. The use of increasingly complex CTFs to minimize the within group standard deviation of a set of normal single photon emission tomography brain images is used to demonstrate the improved registration of images from different subjects using CTFs of increasing complexity.     

Medical image registration

Radiological images are increasingly being used in healthcare and medical research. There is, consequently, widespread interest in accurately relating information in the different images for diagnosis, treatment and basic science. This article reviews registration techniques used to solve this problem, and describes the wide variety of applications to which these techniques are applied. Applications of image registration include combining images of the same subject from different modalities, aligning temporal sequences of images to compensate for motion of the subject between scans, image guidance during interventions and aligning images from multiple subjects in cohort studies. Current registration algorithms can, in many cases, automatically register images that are related by a rigid body transformation (i.e. where tissue deformation can be ignored). There has also been substantial progress in non-rigid registration algorithms that can compensate for tissue deformation, or align images from different subjects. Nevertheless many registration problems remain unsolved, and this is likely to continue to be an active field of research in the future.     

Assessment of dose reconstruction errors in image-guided radiation therapy

Dose reconstruction can be used to improve the accuracy of dose evaluation throughout a treatment course. Its working mechanism is based on deformable image registration (DIR). The purpose of this paper is to develop a method to estimate the dose reconstruction error associated with the inaccuracy of DIR algorithms. To reach this goal, we quantified dominant errors in DIR in terms of unbalanced energy (UE), which were compared with the standard displacement error (SDE). Their high similarity, characterized by Pearson correlation coefficient, was verified through nine 'demons' registration instances performed within simulated reference frames. Based on the similarity, the dose-warping discrepancy at each voxel was defined as a line integral of the dose gradient within the voxel's neighborhood whose boundary was determined by the voxel's UE value. From this definition, the dose reconstruction error was then calculated at each voxel on nine prostate computed tomography images, obtained from a patient treatment course. The average of the Pearson correlation coefficients between UE and SDE over the simulated registration instances was above 70%. The mean value of the dose reconstruction errors in a target volume was calculated for each of nine treatment fractions. The averaged percentage of these mean values with respect to the prescribed dose on the target volume was 1.68%. These results are consistent with contour-based mean dose error evaluations. This paper has established a relation between a registration error and its induced dose reconstruction discrepancy. It allows an automatic validation method to be developed to estimate the dose accumulation error at each voxel in clinical settings.     

Computer-assisted quantification of lung tumors in respiratory gated PET/CT images: phantom study

A computer-aided method was developed to automatically localize tumors in lung PET images of discrete bins within the breathing cycle, followed by an algorithm that registers all the information of a complete respiratory cycle into a single reference bin. Four registration/integration algorithms: Centroid Based, Intensity Based, Rigid Body, and Optical Flow registration were compared as well as two registration schemes: Direct scheme and Successive scheme. Validation was demonstrated by conducting experiments with the computerized 4D NCAT phantom and with a dynamic lung–chest phantom imaged using a GE PET/CT System. Iterations were conducted on different size simulated tumors. Static tumors without respiratory motion were used as gold standard; quantitative results were compared with respect to tumor activity concentration, cross-correlation coefficient, relative noise level, and computation time. After motion correction, the best compromise between short PET scan time and reduced image noise can be achieved, while quantification and clinical analysis become faster and more precise.     

A finite element method to correct deformable image registration errors in low-contrast regions

Image-guided adaptive radiotherapy requires deformable image registration to map radiation dose back and forth between images. The purpose of this study is to develop a novel method to improve the accuracy of an intensity-based image registration algorithm in low-contrast regions. A computational framework has been developed in this study to improve the quality of the 'demons' registration. For each voxel in the registration's target image, the standard deviation of image intensity in a neighborhood of this voxel was calculated. A mask for high-contrast regions was generated based on their standard deviations. In the masked regions, a tetrahedral mesh was refined recursively so that a sufficient number of tetrahedral nodes in these regions can be selected as driving nodes. An elastic system driven by the displacements of the selected nodes was formulated using a finite element method (FEM) and implemented on the refined mesh. The displacements of these driving nodes were generated with the 'demons' algorithm. The solution of the system was derived using a conjugated gradient method, and interpolated to generate a displacement vector field for the registered images. The FEM correction method was compared with the 'demons' algorithm on the computed tomography (CT) images of lung and prostate patients. The performance of the FEM correction relating to the 'demons' registration was analyzed based on the physical property of their deformation maps, and quantitatively evaluated through a benchmark model developed specifically for this study. Compared to the benchmark model, the 'demons' registration has the maximum error of 1.2 cm, which can be corrected by the FEM to 0.4 cm, and the average error of the 'demons' registration is reduced from 0.17 to 0.11 cm. For the CT images of lung and prostate patients, the deformation maps generated by the 'demons' algorithm were found unrealistic at several places. In these places, the displacement differences between the 'demons' registrations and their FEM corrections were found in the range of 0.4 and 1.1 cm. The mesh refinement and FEM simulation were implemented in a single thread application which requires about 45 min of computation time on a 2.6 GHz computer. This study has demonstrated that the FEM can be integrated with intensity-based image registration algorithms to improve their registration accuracy, especially in low-contrast regions.     

基于分段B样条形变配准方法的头颈部伪CT生成研究

目的:探讨分段B样条形变配准方法在头颈部伪CT(sCT)生成中的应用,以及对sCT生成精度的影响。方法:收集已经进行调强放射治疗的鼻咽癌患者45例,每例计划均包括头颈部T1加权核磁共振成像(MRI)和CT图像。使用3D Slicer软件对MRI和CT图像分别进行分段B样条形变配准、整体B样条形变配准、分段刚性配准和整体刚性配准4种方法配准,比较配准后的MRI图像和真实CT图像的Dice相似性系数(DSC)值。随机选取其中的30例患者作为训练集,15例患者为测试集,将配准后的MRI和CT图像通过pix2pix网络进行模型训练生成sCT,对生成的sCT和真实CT进行平均绝对误差(MAE)、结构相似性系数(SSIM)和峰值信噪比(PSNR)值的比较,分析通过阈值法分割为不同组织(骨头、软组织、空气和脂肪)的MAE值。结果:配准后的MRI和真实CT图像比较,分段B样条形变配准方法的DSC值最优;使用4种配准方法生成的sCT和真实CT图像进行MAE、SSIM和PSNR值比较,分段配准方法比整体配准方法好,B样条形变配准方法比刚性配准方法好。分段B样条形变配准方法的MAE值为(74.783±9.8...     

A fast inverse consistent deformable image registration method based on symmetric optical flow computation

Deformable image registration is widely used in various radiation therapy applications including daily treatment planning adaptation to map planned tissue or dose to changing anatomy. In this work, a simple and efficient inverse consistency deformable registration method is proposed with aims of higher registration accuracy and faster convergence speed. Instead of registering image I to a second image J, the two images are symmetrically deformed toward one another in multiple passes, until both deformed images are matched and correct registration is therefore achieved. In each pass, a delta motion field is computed by minimizing a symmetric optical flow system cost function using modified optical flow algorithms. The images are then further deformed with the delta motion field in the positive and negative directions respectively, and then used for the next pass. The magnitude of the delta motion field is forced to be less than 0.4 voxel for every pass in order to guarantee smoothness and invertibility for the two overall motion fields that are accumulating the delta motion fields in both positive and negative directions, respectively. The final motion fields to register the original images I and J, in either direction, are calculated by inverting one overall motion field and combining the inversion result with the other overall motion field. The final motion fields are inversely consistent and this is ensured by the symmetric way that registration is carried out. The proposed method is demonstrated with phantom images, artificially deformed patient images and 4D-CT images. Our results suggest that the proposed method is able to improve the overall accuracy (reducing registration error by 30% or more, compared to the original and inversely inconsistent optical flow algorithms), reduce the inverse consistency error (by 95% or more) and increase the convergence rate (by 100% or more). The overall computation speed may slightly decrease, or increase in most cases because the new method converges faster. Compared to previously reported inverse consistency algorithms, the proposed method is simpler, easier to implement and more efficient.     

A hybrid strategy to integrate surface-based and mutual-information-based methods for co-registering brain SPECT and MR images

Co-registration of brain SPECT and MR images has been used extensively in clinical applications. The complementary features of two major co-registration methods—surface- and mutual-information-based (MI-based)—motivated us to study a hybrid-based scheme that uses the surface-based method to achieve a quick alignment, followed by the MI-based method for fine tuning. Computer simulations were conducted to evaluate the accuracy and robustness of surface-, MI-, and hybrid-based registration methods by designing different levels of noise and mismatch in the registration experiments. Results demonstrated that the hybrid surface-MI-based scheme outperforms both the surface- and MI-based methods in providing superior accuracy and success rates. Specifically, the translational and rotational errors were no more than 1 mm and 2°, respectively, with consistent success rates over 98%. Besides, the hybrid-based method saved 12–53% of the computation efforts, compared with using the MI-based method alone. We recommend the use of hybrid-based method when the orientational differences between the floating and reference images exceed 10°.     

Investigation of four-dimensional computed tomography-based pulmonary ventilation imaging in patients with emphysematous lung regions

A pulmonary ventilation imaging technique based on four-dimensional (4D) computed tomography (CT) has advantages over existing techniques. However, physiologically accurate 4D-CT ventilation imaging has not been achieved in patients. The purpose of this study was to evaluate 4D-CT ventilation imaging by correlating ventilation with emphysema. Emphysematous lung regions are less ventilated and can be used as surrogates for low ventilation. We tested the hypothesis: 4D-CT ventilation in emphysematous lung regions is significantly lower than in non-emphysematous regions. Four-dimensional CT ventilation images were created for 12 patients with emphysematous lung regions as observed on CT, using a total of four combinations of two deformable image registration (DIR) algorithms: surface-based (DIR(sur)) and volumetric (DIR(vol)), and two metrics: Hounsfield unit (HU) change (V(HU)) and Jacobian determinant of deformation (V(Jac)), yielding four ventilation image sets per patient. Emphysematous lung regions were detected by density masking. We tested our hypothesis using the one-tailed t-test. Visually, different DIR algorithms and metrics yielded spatially variant 4D-CT ventilation images. The mean ventilation values in emphysematous lung regions were consistently lower than in non-emphysematous regions for all the combinations of DIR algorithms and metrics. V(HU) resulted in statistically significant differences for both DIR(sur) (0.14 ± 0.14 versus 0.29 ± 0.16, p = 0.01) and DIR(vol) (0.13 ± 0.13 versus 0.27 ± 0.15, p < 0.01). However, V(Jac) resulted in non-significant differences for both DIR(sur) (0.15 ± 0.07 versus 0.17 ± 0.08, p = 0.20) and DIR(vol) (0.17 ± 0.08 versus 0.19 ± 0.09, p = 0.30). This study demonstrated the strong correlation between the HU-based 4D-CT ventilation and emphysema, which indicates the potential for HU-based 4D-CT ventilation imaging to achieve high physiologic accuracy. A further study is needed to confirm these results.     

Computer-aided oral implantology: methods and applications

Aiming at transferring the preoperative planning information to the patient in oral implantology, this study presents an image guided oral implant system (IGOIS), which integrates 3D medical modelling, preoperative surgical planning and intraoperative tracking into a dental navigation system. With the fabrication of a tooth-supported polymer resin template, a non-invasive point-based registration method through ICP algorithm is explicitly discussed. The experimental test shows that deviation between planned and achieved position of the implant is 1.2 mm at the tip and 1.3 mm at the head, which thereby demonstrates that IGOIS can reach a level of accuracy where further clinical developments are feasible.