dPET acquisition captures the spatial distribution of the captured radiotracer changes over time. The dynamic scanning in the aFOV of conventional PET/CT 15 to 25 cm is mainly performed in two ways: step-and-shoot scan mode (SS) and continuous bed motion (CBM). In SS, the number of beds is fixed, and scans must be selected to fit within these beds, with adjacent beds having overlapping sections that can cause partial volume effects. In contrast, CBM does not have fixed beds, allowing for customizable scanning ranges. The absence of bed overlap reduces artifacts and enhances patient comfort, leading to shorter acquisition times and improved efficiency (6).

Conventional PET scanners cannot concurrently capture dynamic data at the far end of the body because of the limits of the short-axis field of view. Therefore, increased axial coverage in a whole-body PET scanner may achieve kinetic analysis on lesions outside of the traditional aFOV. This eliminates time gaps and improves sensitivity, thus eliminating the need for arterial blood collection, which can lead to smaller doses of medication. A new generation of PET scanners entered the market in late 2018, leveraging the most recent advancements in electronics and materials research to attain scanning fields of 1 to 2 meters. These scanners can complete a whole-body acquisition in less than 60 seconds, providing real-time insight into pathophysiologic processes ( Figure 1 ).

SUV is the most commonly used quantitative indicator of radiopharmaceutical uptake. SUV is a simple semi-quantitative analytical technique obtained by sketching the concentration of activity in a lesion after a period of rest from injected radiopharmaceuticals. K_i images and DV images that have been kinetically modeled have better contrast compared to SUV images. Compartmental models are frequently used in the clinic for the evaluation of kinetic models, and the most commonly used compartmental models in PET are the two-tissue compartment model and the three-tissue compartment model, which describes the exchange of radiopharmaceuticals between the blood and tissue compartments, where the four transport rates are represented by the meanings of K₁ related to the inflow of tracer from the blood compartment into the tissue compartments, K₂ related to the efflux, K₃ reflecting the rate of phosphorylation, and K₄ reflects the rate of dephosphorylation. Whereas the parameter that allows the estimation of partial blood volume, also known as vascular density (V_b). V_b and K₁ are correlated and are usually higher than the rate of phosphorylation(K₃).This model differs from the model that does not consider K₄ and V_b, as proposed by Sokoloff et al. The absence of K₄ and V_b results in different values of K₁ and K₃, as K₁ is dependent on V_b, whereas K₃ is dependent on K₄. In addition, the dephosphorylation rate (K₄) of ¹⁸F-FDG may be low but cannot be ignored. Compartmental models have the advantage of directly visualizing different kinetic parameters, but it is computationally intensive and sensitive to noise. As an alternative to compartmental modeling, the Patlak model is a linear graphical method that can approximate K_i using the slope of the graphical curves of the blood input function and the tissue time-activity curve (TAC), and obtains intercepted images related to the volume fraction and the volume of distribution. For parametric imaging, the Patlak model has the advantages of computational efficiency and noise robustness. The associated PET kinetic modeling and parametric imaging flow is shown in Figure 2 .

Kinetic modeling analysis of ¹⁸F-FDG using PET/CT requires AIF based on TAC image-derived AIF. Although arterial blood collection is the gold standard for obtaining AIF (21, 22), it is not the clinically preferred method due to its invasive nature during the procedure (23). Image-derived input function (IDIF) is currently the most commonly used input function. IDIF allows the selection of a dynamic scan of the initial blood pool (heart or aorta) in place of arterial blood collection. The ascending aorta is widely chosen to obtain IDIF because of the strong correlation between the results of the ascending aorta and arterial sampling. Besides, a relatively large ROI can be determined when using an ascending aorta with better statistical properties and less interference from adjacent myocardial spillovers. However, the accuracy of using ascending aorta is affected by body motion and partial volume effects (23). Zhu et al. used IDIF from small vascular regions (carotid artery) as a merging kernel of a priori information, stabilizing the performance of the simultaneous estimation of the input function method and significantly enhancing the accuracy of evaluating early time points of the input function (24). Besides IDIF, a template can be generated using a population-based input function (PBIF). PBIF can be obtained by averaging the arterial blood data collected from the subjects, predetermining the model parameters, and applying a scaling factor for normalization for each subject. Mika et al. compared PBIFs normalized by different scaling factors and found good accuracy and precision using a scaling factor consisting of the area under the curve of the time window and the initial volume of distribution of the developer (22). However, using a single model with a single input function from a single blood pool can lead to inaccurate kinetic parameter estimates for organs with dual blood supplies (liver and lungs) (25). Wang et al. proposed that dynamics can be modeled using an optimally derived dual-blood input function (DBIF). Compared with the traditional single-blood input function (SBIF) and PBIF, DBIF can significantly improve the fitting of liver time-activity curves without the need to invade the arteries for blood sampling or delineate additional regions of interest. Besides, the optimized model demonstrated that FDG blood-hepatic transit rate K_i is significantly correlated with the histological grading of liver inflammation, making it suitable for individual patients (26). Similarly, Yiran Wang et al. used high-temporal resolution dynamic imaging and demonstrated that the DBIF model can be used to better quantify the kinetic parameters and improve the quality of the fit of the efficiency curves under the influence of dual blood supply to the lungs. They also showed that this effect is particularly pronounced in lung tumors (27). In addition, no input function is required in the non-compartmental model to reflect the changes in tumor heterogeneity. The temporal activity data of the tracer in a single voxel within the ROI can be evaluated to show the non-homogeneity of the distribution of the tracer activity over time within the ROI, which is closely correlated with the pathological findings (5, 28, 29). It is also possible to acquire image IDIFs from multiple blood pools (e.g., ventricles and arteries) simultaneously by expanding the axial scanning field of view, such as with long axial PET/CT, and selecting the most relevant vascular IDIFs for kinetic modeling (30).

Although dynamic scanning allows continuous tracking of changes in tracer activity, it is associated with long scanning times, poor patient tolerance, and limited clinical application. The ultra-high-sensitivity whole-body PET scanners can shorten the time for dynamic acquisition. Feng et al. demonstrated that ultra-high sensitivity whole-body PET scanner can achieve whole-body parametric image reconstruction at an early stage of scanning (within 2 minutes of injection) (31). Wu et al. also demonstrated that a nonlinear estimation method can reduce the acquisition time of FDG PET K_i parametric imaging (32). However, Wu et al. also indicated that the shorter imaging time (10 min) is associated with a significantly increased computational cost. Several researchers have used simplified dynamic scanning protocols to effectively shorten the dynamic acquisition time. Compared with the 0-75 min standard parametric images, Hui Wang et al. found that using two short-duration dynamic multiparametric images of 0-6 min (input function) and 60-75 min (equilibrium activity) generated from 0-75 min standard parametric images provide good image quality. Dual time-point imaging can effectively improve gastrointestinal motility artifacts in standard parametric images due to long scanning times by leaving the scanner in the middle of the scanning process (33). Meanwhile, this needs the use of regions of interest from both aortas in both scans to generate the full IDIF and two CT scans to attenuate the two PET images for correction, increasing the radiation dose. Alternatively, PBIF can be used to scale the IDIF extracted from the later time frames of the scan, and a scaling curve can be used to generate the Patlak K_i values. However, the shape of the input curve must be assumed to be the same for each patient. Hasan Sari et al. explored the effect of different scanning cycles on the accuracy of sPBIF and found that sPBIF generated from 55-65 min of image data can accurately assess K_i (deviation <5%) and provide good image quality when it is used for K_i images generated from 20 min (45-65 min post-injection) dynamic scanning (34). Alternatively, a global scaling factor can be used to normalize the Patlak slopes obtained from the data in the later frames, thus eliminating the need for PBIF (35). Although some potential inaccuracies may occur, images obtained using post-scanning give better target-to-background ratios than static images, effectively reducing scanning time ( Figure 3 ).

Whole-body dynamic scanning is mainly achieved by SS and CBM scanning modes in conventional PET scanners due to the limitations of the detector axial scanning field. The high sensitivity of the continuous bed motion listing mode is mainly reflected in the provision of continuous short-interval whole-body dynamic scanning. The whole-body dynamic images are acquired by multi-channel and multibed PET acquisition. Motoki Tomohito Kaji et al. found that the use of CBM scanning mode, which removes the frames with motion in the images, can improve the quality of image acquisition for patients who may move during the scanning process (36). In addition, it is sometimes difficult to identify lesions in the urinary tract and gastrointestinal tract using conventional static PET imaging. As a result, radiologic technicians are needed to distinguish whether the gastrointestinal tract is pathologically ingested or physiologically ingested through delayed imaging. Besides, Motoki Nishimura et al. found that continuous dynamic whole-body imaging can differentiate between pathologic and physiologic uptake based on changes in uptake shape, minimizing the need for such delayed imaging and reducing radiation dose (36, 37). Although multibed, multichannel approaches can provide dynamic imaging with extended AFOV, their temporal resolution is limited by bed motion and scanner sensitivity, which cannot rapidly capture tracer kinetics. In contrast, long-axis field-of-view PET/CT permits dynamic imaging of the entire body at high temporal resolution, acquisitions are no longer segmented by bed position. The long-axis field-of-view PET/CT eliminates the need for segmenting bed position during acquisitions. Besides, long-axis field-of-view PET/CT significantly improves sensitivity by increasing the number of detectors, allowing for dynamic whole-body acquisitions and improved image quality, increasing the detectability of microscopic lesions. Guillaume Fahrni et al. compared SUV and K_i images for 18 different oncologic indications (lesion characterization and staging) and found that K_i can improve sensitivity (from 92.5% to 95%) and accuracy (from 90.24% to 95.12%) compared with SUV imaging, especially for organs with high background uptake (liver). This is because K_i images suppress the non-specific ¹⁸F-FDG signal in the background, resulting in higher contrast in the region of abnormal uptake, improved lesion detectability, and reduced rate of false positives of lesions (7). In addition, the effect of respiratory motion in routine PET scans may lead to inaccurate description (tumor size, location, and morphology) and quantification of small lung nodules near the diaphragm. The use of breath-holding PET technology combined with the counting performance in the long aFOV can minimize motion artifacts and PET/CT image mismatches, especially for small nodules ≤10 mm from the pleura, effectively reducing the false-negative rate of early lung adenocarcinoma (38).

Lung cancer is the leading cause of morbidity and mortality worldwide, according to data released by CA in 2024 (39). Therefore, early detection, accurate diagnosis, and development of individualized treatment plans can improve the survival of lung cancer patients. The dynamic metabolic parameter K_i can be used to improve specificity when differentiating benign and malignant lung nodules. Studies have shown that the K_i parameter is lower in benignancies than in malignancies. Besides, the diagnostic specificity of K_i can be further improved (about 0.830) when the critical value of K_i is 0.0250 ml/g/min, compared with that of SUV_max (40), effectively reducing unnecessary invasive treatment. The K_i value is more sensitive than conventional SUV in identifying early lymph node metastasis. Besides, the use of the K_i and K_i/K₁ values has higher diagnostic specificity in cases where conventional static PET may not accurately identify suspicious lymph nodes (lymph nodes located in the hilar and mediastinal regions), complementing the lower specificity of SUV_max (41).

In addition, dynamic imaging is a valuable non-invasive screening method that can be used to assess treatment effectiveness and predict prognosis. Furthermore, SUV and K_i values are significantly higher in squamous carcinomas than in adenocarcinomas, possibly due to the differentiation of tumor cells combined with GULT-1 and GULT-3 overexpression (42). K_i values are lower in EGFR-expressing adenocarcinoma patients than in patients not expressing EGFR. Also, K_i can improve the differentiation in some NSCLC patients who did not undergo EGFR testing (40). Besides K_i, the tumor load parameters MTV and TLG are reliable overall survival prognostic markers for NSCLC patients receiving platinum-based chemotherapy (43). Notably, the efficacy of a treatment in adjuvant trials is not determined until several years after determining disease-free survival and overall survival. Neoadjuvant trials allow for efficacy endpoints, such as clinical and pathologic response, to be determined in months. Studies have shown that neoadjuvant chemoimmunotherapy can improve long-term survival and increase the chances of cure in patients with unresectable NSCLC. However, there are no suitable predictive imaging markers that can be used to comprehensively characterize the response to neoadjuvant treatment. DaQuan Wang et al. recently demonstrated that the uEXPLORER system can be used for whole-body ¹⁸F-FDG PET/CT dynamic imaging of patients with locally-advanced NSCLC. The results found that Patlak-K_i values were associated with response to induction immunochemotherapy in unresectable locally-advanced NSCLC. Furthermore, they showed that higher FDG K_i values have a better response to induction therapy and a higher level of immune cell infiltration in the primary tumor (44). DaQuan Wang et al. further performed repetitive dynamic scanning of patients during combined immuno-radiotherapy and found that the metabolic profile of the high FDG K_i group after induction immunochemotherapy was significantly reduced during the treatment period. They also showed that the K_i value of the primary foci was significantly correlated with the efficacy and survival of combined immuno-radiotherapy in locally advanced NSCLC patients (45). In summary, dynamic imaging provides non-invasive biomarkers that can predict treatment response in patients with unresectable locally advanced NSCLC.

Hepatocellular carcinoma (HCC) is one of the leading causes of cancer deaths worldwide. Dynamic imaging, dynamic perfusion imaging, and novel imaging agents have emerged as complementary techniques for HCC diagnosis. The blood input function is crucial for precise analysis in dPET kinetic modeling. Early studies used SBIF from the hepatic artery or portal vein for hepatic kinetic modeling (46). However, recent studies have shown that using DBIF is reasonable because of the dual blood supply to liver tissue from the hepatic artery and portal vein (47). Notably, only a few studies have reported the use of DBIF, and no perfect kinetic modeling has been developed yet. Studies have shown that a reversible dual-tissue compartment model for kinetic modeling using DBIF and optimally derived blood-supply fraction hepatic artery can effectively differentiate malignant lesions from healthy liver tissues. Moreover, the model can be used to further distinguish between lesions of intrahepatic cholangiocarcinoma and HCC (48). Dynamic perfusion imaging focuses on the liver and reveals information about the spatial distribution and temporal dynamics of the tumor through dynamic blood flow parameters, such as time to peak (TTP), blood flow, and hepatic perfusion index starting from the injection of ¹⁸F-FDG. Therefore, dynamic perfusion imaging enables quantitative interpretation of PET data (49). Studies have shown that TTP and perfusion index can be used to distinguish HCC from background liver tissue, significantly improving the detection rate of poorly differentiated and moderately differentiated HCC (49, 50).

Contrast-enhanced CT or magnetic resonance imaging can also be utilized for HCC screening. The two imaging are associated with many advantages, including multiphase enhancement characteristics and straightforward data gathering. However, the accuracy of these methods may be compromised when the lesion size is less than 2 cm (51). Also, the sensitivity of ¹⁸F-FDG in detecting HCC is limited to about 36%-68% (52). This is mainly because increased glucose transporter protein expression and hexokinase activity are found in highly differentiated HCC, reducing FDG accumulation in the lesions (53). Some scholars also found that ⁶⁸Ga-FAPI-04 has a higher sensitivity in detecting intrahepatic HCC lesions than ¹⁸F-FDG, especially for tumors with a diameter of less than 2 cm(diagnostic efficiency: about 68.8%) (54). In addition, ⁶⁸Ga-FAPI-04 showed high sensitivity in identifying liver malignant tumors and metastatic lymph nodes (55, 56).

Breast cancer (BC) has become the most common cancer in women, surpassing lung cancer, with an estimated 2.3 million new cases reported worldwide each year, and is the leading cause of cancer deaths in women (57). It is classified into four subtypes, including Luminal A, Luminal B, HER-2 overexpression, and triple-negative, based on the expression of cell proliferation markers (K_i-67), the estrogen (ER), progesterone (PR), and the HER-2 receptor (58). Lack of ER and PR expression in BC correlates with poor overall prognosis, whereas positivity for human epidermal growth factor receptor 2c-erbB2 and K_i-67 are considered good indicators of the invasiveness and metastastic potential of tumors (59). Kornélia Kajáry et al. conducted dynamic PET/CT imaging on 35 primary breast cancers. Their findings revealed that highly aggressive tumors, characterized by high differentiation, hormone receptor negativity (ER and/or PR), and rapid proliferation (high K_i-67), exhibited increased cellular uptake and phosphorylation of FDG (60), and these kinetic parameters can be used to characterize residual lesions following neoadjuvant therapy in locally advanced BC because they less dependent on pre-treatment FDG uptake as static parameters. Mid-treatment evaluation of changes in glucose delivery (K₁) and metabolic rate (MR_FDG) during neoadjuvant chemotherapy can predict recurrence-free survival (RFS). Specifically, a greater reduction in metabolism-to-perfusion ratios (MR_FDG/peak PE and MR_FDG/peak SER) compared to pretreatment levels is associated with longer RFS (61), suggesting a potential new biomarker for therapeutic response in breast cancer. Axillary lymph node metastasis is a critical prognostic factor in locally advanced breast cancer. Traditionally, lymph node status has been assessed invasively through sentinel lymph node biopsy and axillary lymph node dissection (62). These procedures often lead to edema, numbness, pain, and long-term complications (63). The study demonstrated that the TBR and CNR of MR parametric images generated using 6 min single bed dynamic scan and 64 min dynamic whole body PET scan were superior to SUV images, which significantly improved lesion detectability (14).

Microvessel density and tumor angiogenesis can influence the tumor growth and metastasis (64, 65). In contrast, patients with locally advanced breast cancer often exhibit a mismatch between blood supply and metabolic rate within the tumor. This imbalance, potentially caused by inadequate blood flow and/or elevated tumor metabolism, can create a hypoxic microenvironment linked to treatment resistance. This phenomenon is particularly prevalent in triple-negative breast cancer (65). Methods based on a 2-minute half-life ¹⁵O are effective for perfusion assessment (66), however the use of cyclotrons limits its clinical application. Neree Payan et al. used a first-pass scanning approach with a 2-minute dynamic acquisition centered on the chest immediately after injection of ¹⁸F-FDG, followed by a static acquisition 90 min later, which correlated well with ¹⁵O-water measurements of tumor perfusion. In patients with lymph node involvement, primary tumors exhibited high perfusion and K₁ values (67), which may explain the extensive angiogenesis that may promote cancer metastasis.

Pancreatic cancer is a malignant tumor the primarily originates from pancreatic ductal epithelial and follicular cells, with high degree of malignancy and insidious onset. Among its various subtypes, pancreatic ductal adenocarcinoma cancer (PDAC) is the most common type, accounting for about 90% of the cases (68). Intraductal papillary mucinous neoplasm (IPMN) is the precursor of malignant transformation of PDAC (69). Although many IPMNs are benign, differentiating between these and those with malignant potential is crucial. Matthias Lang et al. demonstrated that dPET parameters (TTP, K₁, K₂, K₃, K₄) can effectively distinguish between high-grade and low-grade IPMNs. With 80% sensitivity and specificity, this approach offers a promising tool for predicting IPMN subtypes and potentially avoiding unnecessary surgery for non-malignant lesions (70). In terms of staging and prognosis of PDCA, the dual imaging with ⁶⁸Ga-FAPI and ¹⁸F-FDG allowed precise stratification of postoperative patients, with 18.4% and 10.2% of patients experiencing a change in staging, respectively, but ⁶⁸Ga-FAPI had a lower specificity in identifying primary lesions (71) that was mainly attributed to the overlapping uptake intensity of primary pancreatic tumors and tumor-induced obstructive pancreatitis of the pancreatic parenchyma. Unfortunately, dynamic ⁶⁸Ga-FAPI-04 PET/CT did not effectively differentiate between PDAC and distal obstructive pancreatitis based on K_i values (72). However, dual time-point imaging with a 3-hour delay successfully distinguished between tumor and inflammation-induced fibrosis (72).

Currently, there are limited studies on the dynamics of PET/CT in pancreatic cancer, especially with the low sensitivity or specificity of ¹⁸F-FDG (73), the development of new imaging agents is imperative, and the use of ⁶⁸Ga-labeled trimerized αvβ6 integrin-selective nonpeptide ⁶⁸Ga-Trivehexin has been shown to be superior in PDAC metastases, especially in cancers with low metabolic conversion and/or high stromal ratios, such as liver (74), brain tissue and retroperitoneum (75). In addition, researchers are actively seeking innovative strategies to enhance pancreatic cancer treatment outcomes. A primary focus of targeted therapy involves identifying new targets through the administration of short-lived, highly radioactive radionuclides. In the study by Tadashi Watabe et al., dPET was employed to evaluate ⁶⁴Cu and ²²⁵Ac to label mice with human pancreatic cancer xenografts using a FAP inhibitor (FAPI) and found that it significantly inhibited tumor growth (76), while Yuwei Liu et al. discovered that administration of ¹⁷⁷Lu-FAPI-46 and ²²⁵Ac-FAPI-46 to PANC-1 xenograft mice decreased the tumor growth. However, ¹⁷⁷Lu-FAPI-46 exhibited a less intense but more prolonged therapeutic effect compared with ²²⁵Ac-FAPI-46 (77). These studies suggest that β-therapy and α-therapy targeting FAP may offer effective treatment of pancreatic cancer, but it is currently in the clinical translational phase and further investigations using dPET are needed to determine the best mix of rapid FAP kinetics and physical radionuclide decay. There is a compelling need to develop combinatorial treatment strategies focused on tumor cell elimination.

Current statistics indicate that prostate neoplasm has surpassed lung cancer as the most common tumor in men worldwide and is the leading cause of cancer deaths in men (57). Since 2020, when ⁶⁸Ga-PSMA-11 was approved by the FDA (78), prostate-specific membrane antigen (PSMA)-targeted PET imaging of prostate cancer employing ⁶⁸Ga-labelled complexes is increasingly being adopted in routine clinical treatment worldwide. Studies have shown that the kinetic behavior of ⁶⁸Ga-PSMA-11 varies across different organs and lesions. The irreversible two-tissue compartmental (2T4k) model is more suitable for describing the kinetics in pathological lesions, while the reversible two-tissue compartmental (2T4k) model better fits normal tissue kinetics. In terms of kinetic metrics, the K₃ value effectively distinguished between abnormal lesions and healthy organs with increased absorption of ⁶⁸Ga-PSMA-11, and is expected to gain broader clinical application as an imaging biomarker for differentiating pathological and non-specific ⁶⁸Ga-PSMA-11 uptake in prostate cancer patients (79). Currently, delayed imaging with ⁶⁸Ga-PSMA PET/CT can improve the identification of non-specific PSMA ingestion in patients with prostate cancer (80). In addition, K_i images collected using whole-body PET/CT scanning for 60 min are less noisy and have sharper lesions than SUV images, making them ideal for early detection of small lesions (81). The lengthy 60-min scan time can lead to significant bladder activity, hindering the detection of small lesions like local recurrences or lymph node metastases near the bladder. A combination of a 6-min early dynamic imaging and static imaging protocol (91), as proposed by Jun Wen et al., effectively reduces bladder interference and shortens the PET/CT acquisition time to 45 min. While the improvement in image quality is modest, it enhances patient comfort (82).

The longer half-life of ¹⁸F compared to ⁶⁸Ga offers great convenience in terms of transportation and can be used in delayed imaging protocols. Evidence indicates that ¹⁸F-DCFPyL and ⁶⁸Ga-PSMA-11 exhibit similar biodistribution patterns including the bladder uptake, and that ¹⁸F-DCFPyL has a higher overall detection rate although it did not show a higher detection rate in terms of localized foci (83), this may be due to the better spatial resolution of the images provided by ¹⁸F. A 22-min dynamic ¹⁸F-DCFPyL PET scan effectively detected dominant intraprostatic lesions. Kinetic parameters K_i and K₄ derived from this scan accurately differentiated between tumor and benign tissue (84). These investigations suggest that ¹⁸F-DCFPyL is the most favorable substitute for ⁶⁸Ga-labeled drugs. Moreover, the transition from ⁶⁸Ga-labeled PSMA-targeted compounds to ¹⁸F-labeled PSMA-targeted compounds is expected to gain broader clinical acceptance.

Hypoxia is a common feature of head and neck cancer and a predictor of poor prognosis following radiotherapy (85). ¹⁸F-Fluoromisonidazole (FMISO) has been extensively investigated as a radiotracer for hypoxic PET with good performance (86). Tumor-to-Blood Ratio and Tumor-to-Muscle Ratios are measurements obtained from static FMISO scans and are frequently employed as substitute biomarkers for tumor hypoxia, depending on the specific moment of image capture. Furthermore, tumor-to-blood ratio and tumor-to-muscle ratios values are influenced by the distribution of FMISO within and among tumors. Consequently, this method may inaccurately assess the presence and severity of hypoxia. Milan Grkovski et al. performed a 30-min single-bed dynamic imaging of 120 patients with head and neck cancer, and found that ¹⁸F-FMISO dPET provided data required for parametric mapping of tumor hypoxia, perfusion, and radiotracer distribution volumes, and K₃ was identified as a direct biomarker of hypoxia-mediated ¹⁸F-FMISO accumulation, and K₃ was not dependent on the volume of distribution and time of acquisition with FMISO (87).

Tumor perfusion is a critical factor influencing drug delivery. Changes in perfusion can serve as a prognostic indicator for the effectiveness of both conventional and targeted therapies, such as anti-angiogenic treatments (88). Studies have reported that tumor perfusion and hypoxia can help to predict response to treatment (89). The 20-min truncated ¹⁸F-FMISO dPET scans, i.e., by the kinetic parameters K₁ and K₃ reacting to tumor hypoxia and perfusion can identify alternative markers of hypoxia and perfusion. Currently, two hypotheses have been proposed for the relationship between K₁ and K₃, one is a negative correlation, whereby hypoxia occurs in a hypoperfused tumor volume with a disorganized and dysfunctional microvascular structure. It is hypothesized that the increased separation between tumor cells and functional blood vessels results in diminished oxygen delivery to distal cells, causing diffusion-limited hypoxia. The other is a positive correlation, which implies that areas with good perfusion may experience hypoxia (89). This may reflect the various forms of hypoxia (patchy, banded, and mixed) described in head and neck tumors by Ljungkvist et al, who reported that hypoxia and perfusion can simultaneously occur in very close proximity under the microscope (90). Notably, hypoxia may promote angiogenesis, hyperperfusion and perfusion heterogeneity.