Basic Principles and New Advances in Kidney Imaging

Abstract

Over the past few years, clinical renal imaging has seen great advances, allowing assessments of kidney structure and morphology, perfusion, function and metabolism, oxygenation, as well as microstructure and interstitium. Medical imaging is becoming increasingly important in the evaluation of kidney physiology and pathophysiology, showing promise in management of patients with renal disease, in particular with regard to diagnosis, classification, and prediction of disease development and progression, monitoring response to therapy, detection of drug toxicity, and patient selection for clinical trials.

A variety of imaging modalities, ranging from routine to advanced tools, are currently available to probe the kidney both spatially and temporally, particularly ultrasonography, computed tomography, positron emission tomography, renal scintigraphy, and multiparametric magnetic resonance imaging. Since the range is broad and varied, kidney imaging techniques should be chosen based on the clinical question and specific underlying pathological mechanism, considering contraindications and possible adverse effects. Integration of different modalities providing complementary information will likely bestow the greatest insight into renal pathophysiology.

This review aims to highlight major recent advances in key tools currently available or potentially relevant for clinical kidney imaging, with a focus on non-oncological applications. The review also outlines the context of use, limitations, and advantages of different techniques, and finally emphasizes gaps for future development and clinical adoption.

Introduction

Despite the complexity and heterogeneity of renal structure and function, clinical management of kidney diseases is mainly based on relatively crude laboratory tests. To obtain greater understanding of disease mechanisms and stages, high-resolution imaging techniques have been developed to probe the kidney both spatially and temporally. These tools have subsequently evolved into an essential part of evaluation of kidney physiology and pathophysiology. The advent of tomographic imaging has been particularly pivotal, enabling noninvasive high-resolution discrimination of intra-renal compartments (including cortex, medulla, and collecting system) rapidly enough to monitor functional processes, and with versatility that leverages and matches the rapidly growing understanding of pathophysiological mechanisms.

Several fundamental tools are used in clinical kidney imaging. Given its low cost and availability, despite limited resolution and operator-dependence, conventional ultrasonography has become a routine tool in renal disease. Contrast-enhanced (CE) ultrasound (CEUS), using inert gas microbubbles as contrast agents to investigate microvascular perfusion with reasonably good spatial resolution¹, and ultrasound elastography (UE), enabling in-vivo evaluation of mechanical properties of soft tissue, have introduced novel applications based on ultrasound wave propagation^2,3.

Over the past decade, magnetic resonance imaging (MRI) has emerged as a promising technique for characterization of evolving renal pathophysiology⁴. Structural and functional MRI can be performed in a single multiparametric session, enabling investigation of kidney structure, microstructure, and functional heterogeneity. With notable versatility and no need for contrast agents or ionizing radiation, it is well-suited to serial applications, even in patients with impaired renal function. Its limitations include cost, prolonged scanning, and several clinical contraindications, like metal implants or devices.

Renal scintigraphy is a nuclear medicine technique using radiolabelling to provide combined functional and anatomical information. Depending on the radiopharmaceutical used, it allows both static and dynamic imaging of the kidney. Its limitations include limited spatial resolution, long acquisition time, a high radiation dose and image quality dependence on several factors.⁵ X-ray computed-tomography (CT) offers fast scanning and high spatial and temporal resolution, yet involves ionizing radiation exposure and often requires using iodinated and potentially nephrotoxic contrast material. Positron emission tomography (PET) has low spatial resolution (3–5 mm), uses radiotracers, and is costly and time-consuming, but provides unique metabolic information regarding tissue uptake.

Therefore, appropriate selection of imaging tools can be of enormous value in evaluating the kidney. Using well-defined physiological challenges or molecular targeting further enhances their specificity and sensitivity. Hence, imaging plays a key role in management of patients with renal disease. Medical imaging currently guides a broad range of clinical applications, including diagnosis, classification, and prediction of disease development and progression, monitoring response to therapy, detection of drug toxicity, and patient selection for clinical trials.

This review aims to provide an overview of imaging techniques currently available to probe the kidney (focusing on non-oncological applications), highlight new advances, and emphasize gaps for further development.

Renal structure and morphology

Several imaging-derived aspects of renal anatomy are potentially useful for management of patients with kidney disease.

Renal size.

Changes in kidney dimensions reflect functional deterioration and disease progression. Renal length or volume are useful in evaluation of kidney transplant recipients, and patients with aging, renovascular, or urinary tract diseases, and is central in evaluating progression of polycystic kidney disease (PKD)⁶. Assessment of intra-renal regions might also be revealing; for example, smaller medullary volume in healthy donors predicts subsequent graft failure⁷. Generally, 3D methods like CT and MRI reflect renal dimensions more faithfully than 2D methods like ultrasonography (Figure 1), without relying on geometric assumptions.

Figure 1. Anatomical imaging of the kidney in human subjects.

Top: Coronal 3D Fast Imaging Employing Steady-state Acquisition (FIESTA) anatomical MRI image (A); Co-registered, coronal (contrast-enhanced; CE) 3D CT image (B); and Maximum-Intensity-Projection reconstructed CT image (C) in the same human subject with asymmetric kidneys due to left renal artery stenosis. Both modalities demonstrate clearly a visibly shrunk and less perfused stenotic kidney (red arrow) compared to the contra-lateral kidney. CT also shows aortic calcifications (green arrow). Bottom: B-mode US image (D) and anatomical MR images obtained by T1-weighted (E) and T2-weighted (F) sequences in the same human patient with autosomal dominant polycystic kidney disease. Both US and MRI techniques illustrate the presence of echogenic and fluid-filled cysts in the kidney parenchyma.

Renal roughness.

An irregular cortical surface may indicate nephrosclerosis. CE-CT detects roughness of the kidney exterior in older vs. younger kidneys, independent of cortical atrophy, providing a quantitative index of nephrosclerosis in-vivo⁸. However, studies are needed to determine its sensitivity, as this phenotype is likely limited to advanced renal damage.

Adiposity.

Deposition of perirenal adipose tissue is associated with adverse renal and cardiovascular events. Perirenal fat is measurable using CT⁹, MRI, or ultrasound . MRI-quantified renal sinus fat increases with glucose intolerance, possibly linking metabolic disease and chronic kidney disease (CKD)¹⁰. Moreover, B-mode ultrasound-derived perirenal fat negatively correlates with estimated glomerular filtration rate (GFR) in diabetic patients¹¹.

Kidney stones.

In patients with nephrolithiasis, stone size, composition, and location are important for clinical decision-making. CT is considered the most accurate for initial diagnosis, because ultrasonography might miss small stones.¹² Exciting developments in dual-energy and photon-counting detector CT¹³ allow identification of urate and calcium stones, and possibly rarer stone types¹⁴. Developments to increase resolution and decrease radiation exposure are needed to support non-CE-CT as a first-line tool for evaluation of these patients.

Angiography/Urography.

Traditionally performed by 2D X-ray imaging, evaluation of the renal artery and ureter to detect renal vascular or urinary tract issues and for preoperative evaluation of kidney donors can now be effectively achieved by CT and MRI¹⁵. Paramagnetic iron-oxide nanoparticles, like Ferumoxytol, may provide a safer alternative to gadolinium in MR angiography¹⁶. MR urography, illustrating the urinary tract structure and function, is a promising alternative to intravenous urography, especially in the paediatric population¹⁷. Non-CE MRI offers important advantages over CE-CT for renal artery evaluation¹⁸, including elimination of risk and cost associated with contrast administration and suitability for repeated acquisition, yet must be weighed against potential overestimation of the degree of stenosis and relatively long acquisition times.

Renal Perfusion

Cortical, medullary, and total renal blood flow (RBF) and perfusion (RBF/unit tissue) are influenced by renal function, yet their independent regulation encourages direct measurement of renal hemodynamics. In human subjects, para-aminohippurate (PAH) clearance is useful, but provides no information on individual kidneys or their compartments, underestimates RBF, and is affected by glucosuria¹⁹. Scintigraphy, CT, MRI, and PET overcome many such limitations. A well-established method to assess split renal function and outflow conditions is dynamic renal scintigraphy. Both 99mTc-DTPA and 99mTc-MAG3 can assess relative renal allograft perfusion after transplantation²⁰. 99mTc-PAH, owing to its fast kinetics, excretion properties, and high-quality images, shows promise as a substitute for 99mTc-DTPA²¹, and 99mTc-EC scintigraphy for diagnosis of obstructive uro-nephropathy²². Multi-detector CE-CT affords quantifications of total and regional perfusion using indicator-dilution curves in patients with renovascular disease and hypertension^23,24, and GFR can be concurrently quantified as well²⁵.

PET, a reference standard for quantitative cortical perfusion²⁶, also allows molecular imaging²⁷. PET has been successfully used in patients with renovascular disease²⁸, renal allografts²⁹, and CKD³⁰. The combined PET/CT modalities provides detailed morphology in addition to molecular function and is thus particularly powerful for assessments of renal perfusion and metabolism. Its niche appears to reside in renal cancer, but has also been applied in patients with obesity³¹ or hypertension³². Additional studies are needed to test the clinical validity of this technique to improve the management and outcome of patients with CKD.

Ultra-small paramagnetic iron-oxide-based MRI can assess single-kidney blood volume³³, while non-CE MRI offers non-invasive measurements of both RBF and perfusion. Phase-contrast MRI measures blood flow velocity in individual vessels that occupy at least 16-pixels, to curtail partial volume errors³⁴, and allows computing RBF, alongside several derivative hemodynamic parameters. Although not routinely used in patients, several clinical studies show its potential to support diagnosis and monitoring the early-stage of chronic diseases like CKD, renovascular disease, and PKD³⁵. Arterial spin-labelling (ASL) MRI uses magnetic labelling of water protons in arterial blood as an endogenous tracer to generate maps of regional perfusion. ASL, suited for repeated and longitudinal studies, shows clinical potential to detect early renal damage in ischemia-reperfusion injury, acute kidney injury, or renal transplantation, yet remains restricted to research settings³⁶. Last, hyperpolarized ¹²⁹Xe-MRI has been recently proposed to depict human kidney tissue perfusion³⁷.

Color-Doppler ultrasonography is easily accessible, but user-dependent and technically challenging to make accurate measurements. CEUS can assess microvascular perfusion with satisfactory spatial resolution (Figure 2) and was applied in a variety of kidney diseases, including CKD, diabetic nephropathy, acute renal failure, and renal transplantation². CEUS recently detected important reductions in cortical micro-perfusion in patients with moderate CKD, and perfusion increase following lower-salt intake³⁸. Finally, ultrasound superb microvascular imaging³⁹ allows visualizing low-velocity and small-diameter blood vessel flow using advanced Doppler algorithms and filters to suppress noise caused by motion artifacts, and super-resolution imaging⁴⁰ noninvasively detects renal microvascular changes. Recent 3D-ultrasound advances may enable noninvasive and rapid measurement of renal perfusion with no CE⁴¹, but still await clinical translation.

Figure 2. Latest ultrasound developments in the human kidney.

A) Assessment of tissue perfusion by contrast-enhanced ultrasound (CEUS) in a human transplanted kidney. After continuous infusion of the microbubble contrast agent at a constant rate, a steady state microbubble concentration is reached in the kidney. At that time, a short ultrasound pulse with a very high mechanical index (‘flash’) is generated, resulting in almost complete destruction of the contrast agent microbubbles in the imaging plane. Post-flash images are serially acquired, and perfusion is assessed by the post-flash replenishment kinetics of the volume of microbubbles. B) representative superb microvascular imaging (SMI) ultrasound images of the kidney. Using advanced Doppler algorithms and filters, SMI can suppress noise caused by motion artifacts, without removing the weak signal arising from small vessel blood flow, thereby allowing investigation of the renal microvasculature with no need for contrast media. C) ultrasound investigation of a complex kidney cyst, by B-mode, SMI (left) and CEUS (right).

Function and metabolism

GFR.

99mTc-DTPA scintigraphy has been traditionally used to measure relative GFR, despite requiring 2-4-hour protocol and lower accuracy than alternative exogenous GFR measurement techniques. Dynamic CE (DCE)-MRI allows quantifying single-kidney GFR from time-indicator curves. DCE-MRI can assess renal function in patients with renal artery stenosis and urinary obstruction⁴² and in living kidney donors⁴³, but not in transplant recipients, whose anatomical peculiarities confound standardization of arterial input function⁴⁴. Moreover, DCE-MRI faces important challenges, including i) requirement for gadolinium-based agents, linked to adverse effects in patients with impaired renal function, ii) technical challenges like quantification of gadolinium concentration from signal intensity, spatial registration of the dynamic data and tissue segmentation, and iii) lack of standardization required for clinical practice, which inhibited its widespread clinical use. Importantly, however, new generation Group-II gadolinium-based contrast agents show a markedly improved safety profile⁴⁵. CE-CT also provides single-kidney GFR from time-indicator curves²⁵, as iodine concentration is linearly-related to signal, but the need for contrast agents and radiation exposure restrict clinical application.

Tubular function.

99mTc-MAG3 scintigraphy allows studying tubular function and diagnosing tubular necrosis. Blood Oxygenation Level Dependent (BOLD) MRI, measuring tissue deoxyhemoglobin levels, allows assessing tubular function in-vivo but necessitates a tubule-specific pharmacological maneuver. Indeed, acute changes in oxygenation occur in humans in response to furosemide, and reduced response might reflect tubular damage⁴⁶. Because furosemide does not affect RBF in healthy volunteers, it may help detect changes in renal oxygen consumption regardless of supply⁴⁷,

Solute transport.

Quantitative sodium (23Na)-MRI provides noninvasive quantification of tissue sodium concentration and corticomedullary gradient. Corticomedullary sodium gradient falls in kidney disease and grafts, whereas tissue sodium concentrations increase in diabetic, chronic, and acute kidney injury⁴⁸. Although 23Na-MRI currently requires expensive hardware and software not routinely available on clinical MRI systems, it may become a noninvasive biomarker of physiological renal function and viability. Moreover, the ability to also assess potassium⁴⁹ opens new perspectives in renal physiological imaging.

Metabolism.

PET-derived uptake of [18F]fluorodeoxyglucose (FDG) is a powerful index of kidney metabolism⁵⁰, and ¹⁸F-BCPP-BF uptake detects mitochondrial function in-vivo²⁷, but both are limited by radioactivity and low spatial resolution. Hyperpolarized-carbon-13-MRI is an emerging technique based on enzymatic conversion of injected hyperpolarized substrates, which provides both chemical and spatial information. Endogenous probes like pyruvate and urea, with no adverse effects, are particularly promising⁵¹. Hyperpolarized-MRI kidney studies have primarily focused on hypoxia or oxidative stress, potential unifying mechanisms for kidney disease. Recent clinical translation in oncology, in heart, liver, and brain disease⁵² sets the stage for advances needed to turn hyperpolarized-MRI into a clinical tool.

Oxygenation

Oxygen consumption.

BOLD-MRI is the best-established imaging technique to monitor tissue oxygenation in humans, providing an indirect measure of partial oxygen pressure (pO2). BOLD-MRI may predict renal function decline, and provides powerful insight into the effects of drugs on kidney oxygenation⁵³. In patients with asymmetric disease, single-kidney studies are particularly useful. Despite validation against direct measurement of tissue pO2 in animals, biological validation of BOLD-MRI in humans is warranted. Moreover, many factors including hydration, dietary salt intake, bowel gas, and carbogen and oxygen breathing could potentially affect BOLD-MRI and should be adequately standardized⁵³. Concurrent measurements of RBF or perfusion may assist in result interpretation.

Ischemia.

Diffusion-weighted imaging (DWI)-MRI is capable of estimating ischemia/reperfusion injury delaying graft function⁵⁴, but not predicting renal artery revascularization success⁵⁵. Very recent developments include hyperpolarized-[1-¹³C]pyruvate imaging, allowing to evaluate metabolic status directly in ischemic kidneys⁵⁶ and MRI-CEST (chemical-exchange-saturation-transfer) pH-mapping to evaluate both acid-base homeostasis and renal filtration⁵⁷.

Viability.

Dynamic manganese-enhanced-MRI (MEMRI) allows tracing manganese uptake reflecting cellular viability. In ischemic mouse kidneys, MEMRI revealed decreased cellular viability⁵⁸. This technique may afford noninvasive evaluation of renal viability, but is limited by toxicity that prohibits clinical translation.

Microstructure

Fibrosis.

Fibrosis can be assessed through its impact on renal functional (water diffusion, tissue hypoxia), structural (texture, contour), mechanical (stiffness), and molecular (elastin, macromolecule content) properties. Standard B-mode ultrasound is commonly used to detect kidney fibrosis, which increases cortical echogenicity, decreases its thickness, and alters its contour, although these are insensitive and nonspecific for fibrosis. Ultrasound strain-elastography is the first applied examination to measure kidney stiffness. UE has been widely used to assess liver fibrosis, in particular acoustic radiation force impulse (ARFI), measuring shear wave velocity⁵⁹, and shear wave elastography (SWE), and strongly correlates with kidney fibrosis in experimental models⁶⁰. It improves fibrosis assessment in human kidneys, despite complex acoustic characteristics resulting from kidney heterogeneous histological structure and anisotropy⁶¹. Ultrasound-SWE, denoted by lower susceptibility to other factors and higher repeatability, is now the most widely used method, showing promise to investigate renal allografts, histopathology, or contrast-induced nephropathy. Recent exciting developments include real-time elastography⁶². Moreover, CEUS can evaluate tubular atrophy/interstitial fibrosis in IgA nephropathy⁶³. Nevertheless, confounders and fibrosis-independent contributors to kidney stiffness, like RBF and complex tissue structure, remain to be teased out.

MR elastography (MRE)⁶⁴ detects increased stiffness with CKD⁶⁵ and allograft fibrosis⁶⁶, and may outperform UE in predicting graft loss⁶⁷. T1/T2 mapping, susceptibility-weighted imaging, and magnetization-transfer imaging⁶⁸, are still at early stages of validation. To overcome the low specificity of non-CE-MRI, molecular MRI applies gadolinium-based probes specifically characterizing cellular processes⁶⁹, but is yet to enter clinical development⁷⁰. CE-CT affords imaging of renal fibrosis⁷¹, but due to application of ionizing radiation exposure and intra-arterial injections of nanoparticles, is limited to preclinical settings. Last, 99mTc-DMSA scintigraphy allows detecting renal scarring after acute pyelonephritis⁷².

Microstructure.

DWI, by assessing tissue displacement of water molecules, is sensitive to alterations in the renal interstitium, cellular infiltration, or edema, perfusion and water handling in the tubular compartment⁷³. Moreover, diffusion-tensor imaging (DTI) detects microstructural changes depending on the directionality of molecular motion. Magnetic-resonance relaxometry generates pixel-wise parametric maps of T1/T2 relaxation times reflecting specific tissue properties in kidney injury or graft dysfunction⁷⁴, and together with DWI successfully probed histopathologic microstructure in transplant recipients⁷⁵.

Interstitium

Inflammation and edema.

Several new imaging methods and probes detect cellular or molecular markers of kidney inflammation, yet most are still in the preclinical phase. Vascular leakage and edema can be visualized by MRI T1/T2 mapping⁷⁴, but changes in T1 are non-specific. CEUS detects renal inflammation employing microbubbles conjugated with molecules targeting endothelial inflammation markers⁷⁶. The superparamagnetic iron-oxide (SPIO) MRI probe demonstrates kidney infiltration by macrophages⁷⁷, and when conjugated with molecules targeting tissue-bound complement C3 fragments can monitor kidney inflammation in animal models⁶⁹. While some agents have reached regulatory approval, clinical evidence is still needed to translate SPIO-MRI into a clinical tool to probe inflammation.⁷⁸ 99mTc-OKT3 renal scintigraphy, identifying CD3-positive T-cells, may detect early rejection of renal allografts⁷⁹. Finally, a recent study integrated CXCR4-targeted PET to detect leukocyte infiltrates, with MRI to localize them, in patients with complicated urinary tract infections, informing on the source and extension of local infection⁸⁰.

Abscess/infection.

Imaging techniques play a key role in the diagnosis of infectious complications^81,82. Plain radiography illustrates peri-renal gas in emphysematous pyelonephritis or abscess and calcification in end-stage renal tuberculosis. Ultrasound is often used as screening modality and guiding interventions, portraying kidney mobility, enlargement, parenchyma thickening, and corticomedullary differentiation. Yet, CT is the gold-standard technique for diagnosis and assessment of acute pyelonephritis and to resolve uncertain ultrasound findings. MRI is more sensitive and specific than CT, but not commonly used as first-line, because of high cost and lower accessibility. However, it is helpful in pregnancy and patients with contraindication to iodinated contrast^83,84. DWI is more sensitive than sonography to detect infected renal segments during acute pyelonephritis and to differentiate abscesses from cysts. Finally, [18F]FDG PET/CT proved useful in the diagnosis of renal and hepatic cyst infection in patients with PKD⁸⁵.

Unmet targets and future directions

Since the range of renal imaging techniques is broad and varied, the choice should be driven by the clinical question and specific underlying pathological mechanism, considering contraindications and possible adverse effects (Tables 1–2). Integration of different modalities providing complementary information will likely bestow the greatest insight into renal pathophysiology. In this respect, multiparametric MRI (Figure 3) may represent the lowest-hanging fruit^46,86, with MRI-fingerprinting being a promising development⁸⁷. Moreover, PET/MRI, although currently focusing on oncological applications, uniquely combines structural, functional and molecular imaging⁸⁸ (Figure 4) with considerably lower radiation exposure than PET/CT⁸⁹. Tremendous advances made in clinical renal imaging so far pave the way for exciting future developments.

Table 1.

Overview of potential applications of different imaging techniques currently available or potentially relevant for clinical kidney imaging. The most promising ones are highlighted in bold.

	US	MRI	CT	PET	Scintigraphy
Renal structure and morphology
Renal size	✓	✓	✓
Renal roughness			✓
Adiposity	✓	✓	✓
Kidney stones	✓		✓
Renal artery evaluation	✓	✓ (MRA)	✓
Perfusion	✓ (CEUS, SMI, 3D-US)	✓ (PC, ASL, USPIO)	✓ (CE-CT)	✓	✓ (99mTc DTPA, 99mTc MAG3, 99mTc PAH)
Function and metabolism
GFR		✓ (DCE)	✓ (CE-CT)		✓ (99mTc DTPA, 99mTc PAH)
Tubular function		✓ (BOLD)			✓ (99mTc MAG3)
Solute transport		✓ (23-Na, K)
Metabolism		✓ (HP-13C)		✓ (FDG)
Oxygenation
Oxygen consumption		✓ (BOLD)
Ischemia		✓ (DWI, HP-13C, CEST)
Viability		✓ (ME)
Microstructure
Fibrosis	✓ (B-mode, UE, CEUS)	✓ (DWI, MRE, MRR, MTI, mMRI)	✓		✓ (99mTc DMSA)
Microstructure	✓ (CEUS)	✓ (DWI, DTI, MRR)
Stiffness	✓ (UE)	✓ (MRE)
Interstitium
Inflammation and edema	✓ (CEUS)	✓ (MRR, SPIO)		✓ (CXCR4)	✓ (99mTc OKT3)
Abscess/infection	✓	✓	✓	✓ (FDG)

Abbreviations: ASL=arterial spin labelling, BOLD=blood oxygenation level dependent, CEST=chemical exchange saturation transfer, CEUS=contrast-enhanced ultrasound, CT=computed-tomography, CXCR4=chemokine receptor-4, DCE=dynamic contrast-enhanced, DMSA=dimercaptosuccinic acid, DTPA= diethylenetriaminepentaacetic acid, DWI=diffusion-weighted imaging, FDG=fluoro-deoxy-glucose, HP-13C=hyperpolarised carbon-13, K=potassium, MAG3= mercaptoacetyltriglycine, ME=dynamic manganese-enhanced, mMRI=molecular MRI, MRA=magnetic resonance angiography, MRE=magnetic resonance elastography, MRI=magnetic resonance imaging, MRR=magnetic resonance relaxometry, MTI = magnetization transfer imaging, PAH= p-aminohippuric acid, PC=phase-contrast, PET=positron emission tomography, SMI=superb microvascular imaging, SPIO=superparamagnetic iron-oxide enhanced, Tc=Technetium, UE=ultrasound elastography, US=ultrasound, USPIO = ultra-small superparamagnetic iron-oxide.

Table 2.

Limitations and advantages of available renal imaging techniques

	US	MRI	CT	PET	Scintigraphy
Availability	++++	++	+++	+	+++
Radiation exposure	no	no	++	++++	++++
Contrast agent	Gas-filled microbubbles for CEUS only (allows only vascular imaging, due to the microbubbles size). Very rare complement activation-related pseudoallergy	Gd-based contrast media for few sequences only (DCE, mMRI). Very rare allergic reactions. Risk of NSF, older generation agents contraindicated in patients with GFR<30. Gd retention in the brain	Iodinated contrast media for most indications. Rare allergic reaction. Few side-effects. Risk of contrast-induced nephropathy (GFR<60)	Radiopharmaceutical needed (mainly FDG). No known side effects	Radiopharmaceutical needed (mainly 99mTc DTPA, DMSA or MAG3). Rare and usually mild allergic reaction
Radiotracers	no	no	no	++++	++++
Cost	+	+++	++	++++	+++
Acquisition time	++	+++	+	++++	++++
Spatial resolution	+ (US), +++ (CEUS, SMI)	+++	++++	++	++
Temporal resolution	+++ (CEUS)	++ (DCE)	++++	+	+
3D	no	yes	yes	yes	no
Versatility	+++	++++	no	+	+
Clinical contraindications	None	Metal implants or devices (cardiac pacemakers and implantable cardioverter defibrillators, neurostimulators, ferromagnetic brain aneurysm clips), presence of foreign bodies, claustrophobia	Pregnancy, Hyperthyroidism (CE-CT), claustrophobia	Pregnancy claustrophobia	Pregnancy claustrophobia
Patient weight limit	no	++++	+++	+++	+++
Operator-dependence	++++	++	+	+	+
Patient comfort	++++	++	++	+	+

Abbreviations: CE-CT=contrast-enhanced CT, CEUS=contrast-enhanced ultrasound, CT=computed tomography, DCE=dynamic contrast-enhanced, DMSA= dimercaptosuccinic acid, DTPA= diethylenetriaminepentaacetic acid, Gd=Gadolinium, GFR=glomerular filtration rate, MAG3= mercaptoacetyltriglycine, MRI=magnetic resonance imaging, mMRI=molecular MRI, NSF=nephrogenic systemic fibrosis, PET=positron-emission tomography, SMI=superb microvascular imaging, Tc=Technetium, US=ultrasound.

Figure 3. Multi-parametric MRI sequences in the human kidney.

Representative renal MRI maps obtained in the same healthy human subject, depicting: (A) magnetization-transfer ratio (MTR) via magnetization transfer imaging (MTI), which provides an index of fibrosis (green to light yellow is normal; darker yellow to red indicates fibrosis); (B) blood-oxygen level-dependent (BOLD)-MRI R2* parametric map, a reflection of renal oxygenation (blue to light yellow is normal; orange-red indicates renal hypoxia); and (C), an apparent diffusion coefficient (ADC) map obtained by diffusion-weighted imaging (DWI), used to detect changes in renal microstructure and obstruction to water molecule motion (green to light yellow is normal; dark blue indicate lower diffusion and, potentially, fibrosis). This healthy human kidney does not exhibit significant fibrosis. (D) Representative MR elastography (MRE) images of the kidneys acquired in two healthy volunteers, in coronal and axial plane. Middle panels show wave images, while right panels show stiffness maps. Panel D was reprinted from Serai SD, Yin M. MR Elastography of the Abdomen: Basic Concepts. Methods Mol Biol. 2021; 2216:301-323 ⁶⁴, under the terms of Creative Commons Attribution 4.0 (CC-BY-4.0) International license.

Figure 4. Hybrid imaging of the human kidney.

Representative images depicting A) axial contrast-enhanced volumetric interpolated breath-hold examination, B) axial 18 F-fluorodeoxyglucose PET, and C) fused PET/MRI of the kidneys. Co-registration and fusion of PET and MRI uniquely combines high anatomical detail with functional and molecular information. Adapted by permission from Springer Nature Customer Service Centre GmbH: Suarez-Weiss KE, Herold A, Gervais D, et al. Hybrid imaging of the abdomen and pelvis. Der Radiologe. 2020; 60(Suppl 1):80-89 ⁸⁸.

For example, assessment of the renal microcirculation in-vivo would provide insight on progression of renal microvascular disease and regeneration. Super microvascular imaging³⁹ and super-resolution ultrasound microvessel imaging⁴⁰ may be useful for this purpose. Determination of nephron number and glomerular size in living humans in-vivo remains an unmet target, and cationized-ferritin-enhanced MRI remains to be developed for clinical application⁹⁰. The ability to discern intra-renal cellular composition would also illuminate renal pathophysiology and disease mechanisms.

Artificial intelligence is catching on very quickly in imaging. Methods based on deep-learning have provided excellent automatic 3D segmentation of the kidney using deep neural networks⁹¹, enhance ultrasound image quality⁹², and score kidney stones on non-CE CT⁹³. Moreover, texture analysis based on multiple MRI sequences show promise to assess early renal dysfunction⁹⁴. This is just the beginning, and rapid progress is anticipated.

Clinical translation.

Kidney imaging biomarkers can help address the urgent need to improve kidney disease management and clinical trials outcomes. Among all available techniques, the most auspicious might be ultrasound-based due to its non-invasiveness, ease of use, and gains from rapid technological progress. Additionally, various MR modalities are favorable for mechanism-based investigation of kidney tissue, in particular DWI, BOLD, ASL, MTI and 23-Na techniques.

Imaging biomarkers could help tailor therapy founded on in-depth understanding of disease mechanisms, enrich clinical trials, and identify patients suitable for novel treatments. Moreover, imaging biomarkers suited for serial applications could monitor response to treatment.

Despite encouraging research results, several factors may prohibit advanced renal imaging techniques from achieving regulatory qualification and routine clinical use. For example, research studies emphasize analytical inferences based on group differences, whereas clinical practice needs to deliver to individual diagnostic and treatment decisions. This discrepancy might be addressed by applying machine learning methods to make inferences based on individual structural or functional imaging data⁹⁵. While complex, time-consuming, and expertise-demanding data analysis might forbid rapid clinical decision-making, these might be thwarted by implementation of automated, user-friendly software tools. Amplified volume and complexity of imaging data could be circumvented by developing powerful data archiving and software tools. Furthermore, standardization of data acquisition and analysis may avoiding repetition, unlock silos, and thereby consolidate the discovery and validation powers of multiple groups.

In summary, major recent advances and the rapid ongoing progress in key tools for clinical kidney imaging are preparing the grounds for future incorporation of imaging biomarkers in clinical practice. Rapid gains in resolution and machine learning techniques may allow unprecedented insights into kidney physiology and pathology. The imaging biomarker roadmap recently developed for cancer⁹⁶ provides a useful guide for clinical translation. Parallel tracks of technical validation, biological/clinical validation, and assessment of cost-effectiveness; imaging biomarker standardization; and multicenter studies are needed to cross translational gaps.