OncoReach Steerable Stylet

• The OncoReach Steerable Stylet is a tendon- and flexure-driven needle system that provides controlled in situ steering for minimally invasive interventions.
• It utilizes laser-cut superelastic nitinol and tendon-driven ball-joint designs to achieve predictable curvature, effective obstacle avoidance, and minimized tissue trauma.
• Experimental validations reveal low tip errors (<2.5 mm) and efficient force management, highlighting its potential in both robotic and manual clinical applications.

The OncoReach Steerable Stylet is a tendon- or flexure-driven, highly controllable needle system for in situ steering during minimally invasive procedures, including autonomous robotic interventions and manual interstitial brachytherapy. Leveraging laser-cut, superelastic nitinol or articulated ball-joint architectures, the stylet achieves large, predictable curvature within biological tissue or clinical phantoms. Distinguishing itself from conventional, rigid, or classic bevel-tipped needles, the OncoReach platform enables obstacle-avoidant trajectories and lateral access to previously inaccessible targets, while accommodating physiological motion and minimizing tissue trauma (Kuntz et al., 2022, Kheradmand et al., 20 Jan 2026).

1. Mechanical Construction and Actuation

The OncoReach Steerable Stylet encompasses multiple architectures, most notably:

• Laser-Patterned Flexure Design (robotic steering): A superelastic nitinol shaft (outer diameter 0.80 mm, inner lumen 0.50 mm) is laser-cut along a distal segment (L_active ≈ 20 mm) into N ≈ 20 thin flexure hinges, each with length ℓ_h = 0.5 mm, thickness t_h = 50 µm, and width w_h = 600 µm. The hinges act as serial rotational springs (stiffness $K_\theta \approx E I_h / \ell_h$, where $I_h \approx (t_h w_h^3)/12$), generating an aggregate maximum curvature κ_max ≈ 0.10 mm⁻¹ (R_min ≈ 10 mm).
• Tendon-Driven Ball-Joint Design (brachytherapy steering): Four nitinol tendons route through 3D-printed disks containing 0.8 mm stainless steel ball bearings (spherical joints) positioned along a 33 mm section. Tendons, actuated via DC motor and PID-controlled tensioning (setpoints up to ~10–14 N), induce tip deflections up to 11.05 mm (15 ga., 30-tip-joint configuration), with the articulated section's low bending stiffness (K_bt,joints = diag([0.0008, 0.0008, 0.0047]) N·m²) determining compliance (Kheradmand et al., 20 Jan 2026).

Stylets are engineered for compatibility with standard clinical needle gauges (e.g., 15- and 13-gauge ISBT needles), and integrate electromagnetic coils or fiducials for real-time tip localization.

2. Kinematic and Physical Modeling

• Constant Curvature Model: In a homogeneous tissue environment, the stylet’s deformed shape is parameterized by a planar curve of constant curvature κ, with arc-length s:
  • $x(s) = \frac{1}{\kappa}\sin(\kappa s)$, $y(s) = \frac{1}{\kappa}(1 - \cos(\kappa s))$, $\theta(s) = \kappa s$.
  • The harvested κ is bounded by both the stylet’s flexure geometry and the tissue’s mechanical response, typically constrained as $0 ≤ κ ≤ κ_{\rm max}$.
• Cosserat Rod/String Model: For advanced shaft–tissue interaction, the stylet and cannula are modeled as parallel Cosserat rods:
  • Each tube i has material frame $R_i(s)$, linear strain $v_i(s)$, and angular strain $u_i(s)$.
  • Equilibrium and constitutive relations are captured as $\partial_s [n_i; m_i] + [...] + [f_i; \tau_i] = 0$, with force/moment responses intertube-coupled.
  • The shaft-to-tissue contact model defines the normal force per unit length as $f_n(s) = n(s) κ(s)$, and the frictional force per unit length as $f_f(s) = μ(s)[f_c(s) + f_n(s)]$, where n(s) is the internal compressive force, μ(s) the local kinetic friction coefficient, and f_c(s) the tissue’s compressive reaction (Bentley et al., 2021).

Model-experiment validation shows tip-position errors of <2.5 mm for 15–30 joint configurations at maximum tension, corresponding to <1.1% of the length (Kheradmand et al., 20 Jan 2026).

3. Control Architecture and Planning Algorithms

• Closed-Loop Control: Real-time controllers use electromagnetic tracking to close the loop on both insertion depth (δ) and shaft rotation (φ), via dual PI controllers for precise pose tracking.
• Motion Planning:
  • Sample-Based Planning: Preoperative planners (e.g., RRT-CC or RG-RRT) generate candidate paths incorporating anatomical obstacle data from segmentation.
  • Cost Functions:
  • Obstacle-aware length/clearance cost: $J[γ] = α \int_0^L d_{\rm obs}(γ(s))^2 ds + β L$.
  • Bottleneck force cost (force-optimal): $C_F(τ) = \max_{t,s} f_{n,t}(s)$, where f_{n,t}(s) is computed via the Cosserat-string ODE, with tip boundary $n(L) = F_p$.
  • Replanning is triggered when deviations exceed ε_replan (≈2 mm).
• Respiratory Motion Compensation: For moving organs (lung biopsy), insertion is segmented into sub-arcs of Δs = 10 mm, each inserted during a ventilator-gated, peak-inhalation breath hold (T_hold ≈ 10 s), exploiting chest wall fiducials for phase tracking (Kuntz et al., 2022).

4. Force Safety and Tissue Interaction Optimization

• Cosserat-String–Based Force Model: Explicit modeling of tissue–shaft forces is used to minimize lateral shearing risk, a critical factor for clinical safety. The peak normal force along the shaft is computed by integrating shaft equilibrium equations with measured tip forces and calibrated friction parameters (μ, f_c, F_p).
• Bottleneck Cost–Based Planning: Trajectories are selected to cap the worst-case normal (perpendicular) tissue force below operator-specified thresholds. In simulations, this approach reduces peak tissue force by ≈62% versus classic path-length planners, with negligible path-length penalty (≈0.07%) (Bentley et al., 2021).
• Implementation for OncoReach: Parameters are tuned in bench-top gels, ex vivo tissues, and in situ, with automated re-planning if measured forces deviate from predicted values.

5. Experimental Validation and Performance Metrics

Configuration	Max Tip Deflection (mm)	Tip Error at Max Tension (mm)
15 ga., 20 joints (tip)	9.87	2.20 (1.1% length)
13 ga., 20 joints (tip)	6.49	–
15 ga., 30 joints (tip)	11.05	2.33 (1.1%)
15 ga., 20 joints (base)	29.40	11.16 (5.6%)

• Robotic Needle Steering (In Vivo Lung, n=3): Porcine targeting errors: 1.8–3.4 mm, with no observed complications or obstacle collisions. Robot outperformed manual bronchoscopy in ex vivo placements, achieving mean targeting errors of 3.4 ± 3.2 mm (robot) versus 14.7 ± 8.6 mm (manual), p ≈ 1×10⁻³ (Kuntz et al., 2022).
• Brachytherapy Phantom Pilot (N=1 expert): Steered needles reached targets beyond straight-needle access, with tip deflection increases of >6–10 mm in all directions, suggesting improved coverage from less invasive entry points (Kheradmand et al., 20 Jan 2026).

6. Clinical Applications and Implications

• Minimally Invasive Access: By facilitating curved needle/stylet trajectories, OncoReach enables access to otherwise unreachable targets or lateral tumor regions, beneficial in lung and gynecologic oncologic interventions.
• Reduced Procedure Burden: Enhanced trajectory flexibility can reduce the total number of required needle insertions (e.g., in ISBT), minimizing patient trauma and potentially decreasing overall radiation dose to organs at risk (Kheradmand et al., 20 Jan 2026).
• Safety and Conformance: Bottleneck-force–driven planning and closed-loop force monitoring directly address tissue damage risks by constraining peak shaft–tissue interaction forces (Bentley et al., 2021).
• Workflow Integration: The device can be operated both as a robotic, autonomous platform (with real-time motion planning and EM-guidance for lung biopsy) and as a reusable, handheld stylet for expert manual steering (ISBT).

7. Technical Evolution and Modeling Paradigms

The OncoReach Steerable Stylet platform represents an overview of proven continuum mechanics, robotic planning, and clinical device design. Central methodologies include:

• Serial Flexure and Tendon-Driven Architectures: For selectable trade-offs between compliance, steering agility, and axial rigidity.
• Cosserat Rod and String Models: For high-fidelity prediction and optimization of shaft shape and interaction forces, validated against gel phantom and ex vivo experiments.
• Sampling-Based, Cost-Driven Planning: For optimizing both geometric and force safety metrics in complex anatomical scenarios, supporting real-time, adaptive trajectory generation in response to intraoperative imaging and physiological motion (Kuntz et al., 2022, Bentley et al., 2021, Kheradmand et al., 20 Jan 2026).

The integration of mechanical customization (joint count, placement, gauge), advanced modeling (two-tube rod coupling, bottleneck force computation), and pilot clinical validation positions the OncoReach platform as a reference architecture for next-generation steerable needle/stylet interventions in both manual and robotic clinical workflows.