Background
In the 1970s, the first computed tomography (CT) scanner was implemented in the clinical
routine. During the past decades, many innovations have continuously improved CT imaging.
Conventional multi-slice detector CT (MDCT) in clinical routine is typically equipped
with energy-integrating detectors (EID). Images result from indirect conversion of X-ray photons into visible
light and then into an electric signal [1]
[2]
[3]. By integrating the absorbed energy over a short period of time, the information
from each individual X-ray photon’s energy is lost [4].
New CT scanners use photon-counting detectors (PCD). In contrast to EID, PCDs are
able to directly convert X-ray photons into an electric signal using semiconductors
[5]. Based on this groundbreaking technology, PCD-CT can overcome many limitations of
EID-CT. The main advantages of PCD-CT are the elimination of electronic noise, the
improvement of spatial resolution, the intrinsic spectral imaging capabilities, and
the potential reduction of radiation dose [2]
[3]
[5]
[6]
[7]
[8].
PCD-CT was introduced in clinical routine in April 2021. Two PCD systems are currently
used: a commercially available system (Naeotom Alpha, Siemens Healthineers, Erlangen,
Germany) and a clinical prototype (SPCCT, Philips GmbH, Hamburg, Germany). Most studies
that are covered by this review and also our personal experiences focus on the former.
The aim of this review was to summarize the first experimental and clinical experiences
with PCD-CT in musculoskeletal radiology.
Results
Improved spatial resolution and reduction of radiation dose
Due to the ability to directly convert X-ray photons into an electric signal and the
use of smaller detector pixels, PCDs exhibit (ultra-) high resolution capabilities
[9]. Especially in bone imaging, high spatial resolution is important for the imaging
of fractures, bone healing, malignancies, and the visualization of tiny osseous structures
[9]. Previous cadaveric studies have demonstrated the higher spatial resolution of PCD-CT
compared to EID-CT [2]
[3]
[10]
[11]. Recent studies also highlighted the higher spatial resolution even with a lower
radiation dose on a PCD-CT scanner compared to an EID-CT scanner [12]
[13]
[14]
[15]. Regarding specific body regions, previous phantom/mouse and cadaveric studies pointed
out the improved visualization and delineation of tiny bone details in the following
anatomical regions: shoulder [16], wrist [12]
[17], temporal bone and skull base [13]
[15], appendicular skeleton [18], spine [15]
[19]
[20]
[21], paranasal sinus [22], and elbow [23]. Especially ultra-high resolution reconstructions have the ability not only to improve
visualization of lung structures [24]
[25]
[26]
[27]
[28]
[29]
[30] and cardiac structures [31]
[32]
[33]
[34], but also of bones [11]
[12]
[13]
[15]
[16]
[20]
[35]
[36]
[37]
[38]
[39]
[40]
[41].
For example, using a mouse model, intervertebral spaces of the cervical spine were
visible on PCD-CT at a radiation dose of 5 mGy CTDIvol (computed tomography dose index volume), while a dose of 20 mGy CTDIvol was necessary to discriminate different vertebrae on EID-CT. However, delineation
was still unsharp. To clearly visualize the skull base and the inner ear of a mouse,
10 mGy CTDIvol was sufficient on PCD-CT, whereas sharp delineation was not possible on EID-CT even
at 20 mGy CTDIvol
[15]. [Fig. 1] shows an example of ultra-high resolution imaging of the skull base of a mouse on
different CT scanners, highlighting the capability of PCD-CT to provide sharp resolution
of even tiny bone structures. In [Fig. 2] we provide a comparison between PCD-CT and EID-CT at a radiation dose of 20 mGy
CTDIvol. This example highlights the improved spatial resolution and detailed visualization
of small bone structures. Using an ultra-sharp reconstruction kernel (Hr98) on the
PCD-CT scanner, the craniocervical junction, the atlantodental space, and also the
intervertebral disc were clearly visualized in a mouse model. The higher spatial resolution
on PCD-CT was also shown quantitatively; our group assessed the edge sharpness at
the lumbar and cervical spine. On a PCD-CT scanner, edge sharpness was significantly
higher compared to an EID-CT scanner in all analyzed regions [15]. Similar results with higher cortical sharpness were also reported by Sonnow et
al. using a cadaver study with an artificially created elbow fracture [23].
Fig. 1 Ultra-high resolution imaging of the skull base of a mouse on an EID-CT scanner (A: 20-slice EID-CT, B: 64-slice EID-CT), and on a PCD-CT scanner (C). Image C shows the remarkably sharp visualization of tiny structures of the skull base with
a size of only about 2 mm.
Fig. 2 Ultra-high resolution imaging of a mouse in comparison between a 20-slice EID-CT image
(B, D) and PCD-CT image (A, C), both at 20 CTDIvol. Image A shows the sharp delineation of the atlantooccipital joint (red arrow) on
a PCD-CT scanner. In image C, the delineation of the atlantodental space (yellow arrow) and the intervertebral
discs (blue arrow) is highlighted, whereas on an EID-CT image (B, D) the described structures cannot be visualized clearly.
First clinical studies confirmed that spatial resolution is higher on a PCD-CT scanner
compared to an EID-CT scanner in musculoskeletal imaging. For example, Benson et al.
showed improved delineation of temporal bone compared to EID-CT [42], Rajagopal published the first clinical results in skull base imaging [43], while Baffour et al. showed results of imaging of the pelvis and shoulder [36]. Rajendran et al. shared the first results for wrist imaging [35] and Rau et al. showed improved spatial resolution for PCD-CT in spine imaging [44]. A recently published study by Marth et al. showed comparable image quality of the
lumbar spine on PCD-CT (compared to EID-CT) at significantly lower radiation doses
[45].
New image reconstruction techniques including iterative reconstruction can further
improve the spatial resolution of reconstructions in PCD-CT [9]
[29]
[46].
The first clinical studies also addressed the diagnostic performance of PCD-CT for
the detection of malignant bone lesions, e.g., myeloma [47]. Recent studies in patients with myeloma showed similar lesion detection on PCD-CT
compared to EID-CT, however, with a lower radiation dose [48], improved spatial resolution, and visibility on PCD-CT compared to EID-CT [49]
[50]. Wherse et al. reported improved visualization of bone metastases using ultra-high
resolution kernels on an experimental PCD-CT scanner in a case series of breast cancer
patients [51]. [Fig. 3] shows an example of a patient with bone metastases of breast cancer with CT scans
on a PCD-CT scanner and on an EID-CT scanner (time frame about three months between
the two examinations), allowing a direct comparison of both scanners. These images
highlight the improved spatial resolution of PCD-CT for the delineation of critical
findings, e.g., bone metastases.
Fig. 3 Example of a patient with bone metastases from breast cancer and short-term follow-up
on an EID-CT image (A) and a PCD-CT image (B) which allows a direct comparison of both scanners in the delineation of critical
findings such as metastases.
Exploiting the ability of PCD-CT to improve spatial resolution and to reduce image
noise, it has a high potential to reduce patients’ radiation dose. Our first experimental
studies showed the possibility to detect tiny bone structures on PCD-CT with a lower
CTDIvol compared to EID-CT [15]. In musculoskeletal imaging, low-dose CT protocols are important for the detection
of bone lesions, especially in patients with the need for repeated CT imaging [9].
Promising results for radiation dose reduction in musculoskeletal imaging were also
shown in recent clinical studies. Previous studies reported improved visualization
of temporal bone compared to MDCT at lower radiation doses [13]
[41]
[42]
[52]. A similar dose saving potential was also shown for the spine [19]
[44], shoulder and pelvis [36], and the wrist [35]. Other clinical and cadaver studies showed improved visualization without dose penalty
[16]
[23]
[53].
Electronic noise and signal-to-noise ratio
Another advantage of the PCD technology is the elimination of electronic noise due
to the direct conversion of X-ray photons into an electrical signal [1]
[2]
[5]. Consequently, images show an improved signal-to-noise ratio (SNR) as well as an
improved contrast-to-noise ratio (CNR), which has also been shown, e.g., for abdominal
imaging [2]
[5]
[54]
[55]. SNR is defined as the ratio between CT values in a predefined region (e.g., trabecular
bone) and the image noise (standard deviation of CT values). In first experimental
studies, we showed that image noise was significantly lower on PCD-CT compared to
EID-CT at various CTDIvol values with a consequently significantly improved SNR (ratio between CT values in
bone and standard deviation of CT values in air), which resulted in a relative improvement
in SNR of up to 36% [15].
Other PCD-CT studies assessed the CNR, calculated as the ratio between the difference
between CT values in bone and fat and image noise (standard deviation measured in
bone or fat) [17]
[23]. Previous studies showed that CNR is significantly higher in PCD-CT compared to
EID-CT, resulting in an improved delineation and visualization of bone structures
[11]
[16]
[17]
[19]
[23]
[43]
[56].
Spectral imaging
Due to the multi-energy acquisition capability of PCD, all images are acquired in
spectral mode – without adding radiation dose. Therefore, spectral image postprocessing
can be performed for each scan [9]. In bone imaging, material classification (for example, the detection of urate crystals
or calcium pyrophosphate crystals) is an important issue in the clinical routine [9]
[57]. Dual-source dual-energy CT is an established proven technique in order to visualize
urate crystals in patients with suspected gout [58]. However, this often implies a trade-off between using spectral imaging or a high-resolution
technique [9]. In clinical routine and in patients with suspected gout, a combination of high-resolution
and multi-energy imaging would be desirable. Using PCD-CT, both UHR mode and spectral
imaging can be combined. [Fig. 4] shows a case with evidence of a small amount of urate crystals in the metatarsophalangeal
joint and points out the value of PCD-CT for imaging in suspected gout. Spectral imaging
can also be used for imaging of joint spaces and delineation of small cartilage defects
after intraarticular contrast injection [9]
[59]
[60]. A recently published study by Garcelon et al. showed the ability of spectral imaging
to visualize cartilage and subchondral cysts on knee CT without using contrast agent
in cases with osteoarthritis [61].
Fig. 4 Spectral imaging in the diagnostic work-up of suspected gout. Image A shows tiny urate crystals (green) next to the metocarpophalangeal joint. In B, the joint is magnified and shown in sagittal reconstructions. Image B points out
the possibility of PCD-CT to visualize and locate very tiny urate crystals.
Spectral imaging is also important in the diagnosis of acute trauma and in the detection
of fractures. In most cases of suspected fractures, radiography is performed with
additional CT in unclear cases or in cases with a need for further visualization of
fracture morphology. However, there are a few cases (especially in fractures not involving
the bone cortex) in which CT cannot detect the fracture. Then, further MRI is necessary
to detect edema-like signal intensity as an indirect sign of a fracture [62]. However, access to MRI may be limited in clinical routine and in the emergency
setting due to its availability, the long duration of the examination, and potential
contraindications (e.g., pacemaker, metal implants). Previous studies on dual-energy
CT (DECT) delivered promising results in the detection of bone marrow edema using
virtual non-calcium images (VNCa) [63]
[64]
[65]
[66]
[67]
[68]. This virtual spectral postprocessing enables the detection of discrete changes
in the bone marrow by suppressing the high attenuation of bone structures and thereby
enhancing the visualization of water (edema) in the bone marrow. This postprocessing
can be routinely performed in PCD-CT [69].
Preliminary data from our department showed promising results with detection of bone
marrow edema in patients with fractures of the spine ([Fig. 5]). In this case, we showed CT of the thoracic spine of a patient with acute back
pain. Conventional CT images were not able to detect an acute fracture. After postprocessing
of the images using spectral data, we were able to show bone marrow edema (shown in
green) in thoracic vertebrae 6 and 7 but not in the other visualized vertebrae. The
bone marrow edema was confirmed with the gold standard imaging method, MRI, within
24 hours after the CT scan. Postprocessing is based on virtual non-calcium (VNCa)
imaging, a three-element decomposition technique that is known from DECT [69]
[70]. In this technique, the typical attenuation of yellow and red bone marrow is defined
as the baseline. It enables the generation of calcium images and VNCa images separately.
Bone trabeculae are removed, which permits direct visualization of the bone marrow
[69]. Also, for imaging of pelvic fractures, this new technique shows promising results
in our first clinical experiences. Using VNCa, bone marrow edema was also detected
in the pelvis in two examples.
Fig. 5 CT imaging of the thoracic spine in a patient with acute upper back pain. Conventional
CT images show no acute fracture (A); image postprocessing (virtual noncalcium, VNCa) shows bone marrow edema in thoracic
vertebrae 6 and 7 (B). Increases in attenuation (e.g., bone marrow edema) are shown in green, whereas
fat is visualized in purple. Bone marrow edema was confirmed by MRI within 24 hours
(C).
These findings open up new possibilities and might obviate MRI in the near future
in selected cases. Further prospective studies are necessary to evaluate the diagnostic
accuracy of spectral imaging for the detection of bone marrow edema in different regions
of the body and to compare it to MRI as the gold standard imaging method.
Metal artifact reduction
Trauma surgery often requires the use of metal implants in the spine or the extremities.
In the postoperative setting and also during follow-up, further CT imaging is often
necessary. However, due to the formation of metal artifacts, interpretation of adjacent
structures is often not possible. Many patients also have dental implants which often
have strong beam hardening artifacts. This renders the visualization of neighboring
structures nearly impossible. Modern CT scanners employ algorithms for the reduction
of beam hardening artifacts using iterative metal artifact reduction techniques [9]
[71].
However, spectral imaging was also shown to have great potential for metal artifact
reduction. With the multi-energy acquisition of PCD-CT with each scan, spectral postprocessing
can be routinely performed. Spectral postprocessing enables the generation of virtual
monoenergetic images (VMI) [1]
[2]
[5]
[9]. Many previous studies assessed techniques of metal artifact reduction. DECT showed
promising results with a reduction of beam hardening effects at higher keV levels
[9]
[72]. The most convincing results were shown for the combination of iterative metal artifact
reduction (IMAR) and higher keV levels on DECT [73]
[74]
[75]
[76]
[77].
The first phantom-based studies on a PCD-CT scanner assessed the value of IMAR [78], tin filtration [79], VMI [80], projection-based material decomposition [81]
[82]
[83], as well as the combination of IMAR and VMI [84]
[85].
The first clinical studies on PCD-CT also showed promising results for metal artifact
reduction, for IMAR, as well as for VMI and/or the combination of both methods.
A recently published study by Popp et al. showed that VMI enables metal artifact reduction
after spine surgery and determined 110 keV as the optimal energy level for artifact
reduction [86]. As spectral imaging is assessed routinely on a PCD-CT scanner, automatic 110 keV
reconstruction can be easily performed and can help radiologists and clinicians evaluate
CT images after spine surgery ([Fig. 6]). Similar results were also shown for artifact reduction after total hip replacement
[87].
Fig. 6 Patient with screws in the lumbar spine and the pelvis. The left column (A, C) shows conventional 70 keV reconstructions. 110 keV reconstructions show a significant
reduction of metal artifacts both in the spine and in the pelvis (B, D).
For the reduction of dental metal artifacts, IMAR showed promising results. This effect
could further be enhanced with high keV VMIs on a PCD-CT scanner, as recently published
by Risch et al. [88] ([Fig. 7]). The reduction of dental metal artifacts is very important in clinical routine.
It is very common, and it often limits the assessment of cervical structures or structures
in the brain. Other clinical studies also suggested the combination of IMAR and VMI
for metal artifact reduction in the hip [89] and for dental implants [90].
Fig. 7 Contrast-enhanced CT of the neck in a patient with dental implants. A shows conventional 70 keV reconstructions with a Qr60 Q 3 kernel. Using IMAR, there
was a significant reduction of metal artifacts with consequently better delineation
of surrounding structures (B).
Critical assessment
In the following paragraph, we aim to perform a critical assessment of the new technology.
Despite the many advantages that were presented above, there might also be some limitations
in clinical routine and musculoskeletal imaging at this time. Due to the rapid technical
development, there are many different reconstruction algorithms and reconstruction
kernels available. These differences between the centers might lead to confusion.
Therefore, standardization of protocols for musculoskeletal imaging and specific issues
is mandatory.
Due to the possibility of reconstructing spectral data within each scan, huge data
sets are generated. This requires not only significant computational power, but also
requires radiologists to examine the rapidly growing number of images and to write
the reports. To avoid a loss of information, more technical and also human resources
might be necessary. This might be challenging, especially in times of shrinking resources.
Due to the novelty of the PCD technology, large studies and especially multi-center
trials are currently not available. Regarding various clinical scenarios, data about
the sensitivity and specificity of the new reconstruction algorithms or spectral data
sets is lacking. Therefore, many applications (e.g., detection of bone marrow edema)
are undergoing clinical testing at the moment and the diagnostic accuracy must be
demonstrated in clinical trials.
Summary
This review summarizes recent clinical and experimental studies as well as our personal
experiences with a new PCD-CT scanner regarding advances in musculoskeletal imaging.
In summary, this new groundbreaking technology has many advantages compared to conventional
CT: reduction of image noise and elimination of electronic noise, improved spatial
resolution combined with reduction of radiation dose, spectral imaging with the potential
to detect bone marrow edema and gout, as well as metal artifact reduction.
Based on previous studies, PCD-CT is a very promising technology in musculoskeletal
imaging that will be introduced in a wider range of clinical applications and might
have the potential to improve imaging combined with a reduction of radiation dose.
However, further studies are necessary to assess the manifold possibilities of PCD-CT
in musculoskeletal imaging, to examine new reconstruction and postprocessing methods,
and to improve diagnostic accuracy.
